Language Variation and Innovation in Teesside English

Mari Carmen Llamas

Submitted in accordance with the requirements for the degree of Doctor of Philosophy

The University of Leeds School of English

September 2001

The candidate confirms that the work submitted is her own and that appropriate credit has been given where reference has been made to the work of others. Acknowledgements

This study was funded by a White Rose Scholarship through the Universities of Leeds and Sheffield. I gratefully acknowledge the support of the scholarship which allowed me the opportunity to undertake the present research.

1 would like to thank Clive Upton, at the University of Leeds, and John Widdowson, at the University of Sheffield, for acting as supervisors throughout the completion of this research. Particular thanks go to Paul Foulkes and Dominic Watt for constant support and advice. I also wish to thank Lesley Milroy and Judy Dyer for the interest they have shown in the work and the guidance they have provided in interpretation of the results of the study. My thanks also go to Mark J. Jones, Louise Mullany and Ghada Khattab for countless favours.

Many other people are owed acknowledgement and thanks for their assistance throughout the course of the research: these include, Ann Williams, Paul Kerswill, Bridget Anderson, Jason Jones, Anne Fabricius, Paul Johnston, Peter Stockwell, Dave Britain and Jane Stuart-Smith.

The assistance and encouragement provided by Geoff and Jenny Braddy, Val Magee and Jean Christie are also gratefully acknowledged.

Finally, my thanks go to Sarah, Mark, Paddy, Rayees, Beck, Joe, and Helen. Abstract

This thesis presents a study of socially-conditioned phonological variation in a hitherto unresearched urban variety of British English. The variety in question is that of Teesside, on the North East coast of England. A corpus of conversational and word-list data has been collected through the use of an innovative method of data elicitation designed and systematically trialled in the present study for a large-scale project on linguistic variation in Britain. The method combines the elicitation of comparable data on lexical variation with the collection of samples of informal speech on which phonological and, to some extent, grammatical analyses can be undertaken. The data for the present study are taken from a socially homogeneous group of 32 speakers who are divided by gender and grouped into four emically-defmed age cohorts to test for both generational differences suggestive of change in progress in the variety, and fine-grained age differentiations between young speakers. Consonantal variables are investigated and external, speaker-based factors are examined in relation to the levelling out of localised forms and the diffusion of innovatory forms into the variety. Furthermore, the study is an investigation of the interdependence of language and identity. Analyses of speakers’ perceptions of language variation, community identity and cohort identity within the community identity are combined with correlational analyses of the linguistic variables. A language ideology framework is invoked to interpret the linguistic trends uncovered. Findings reveal that the adoption or increased use of forms associated with varieties further south or further north of the locality in question does not appear to affect the local orientation of the speaker. Table of Contents

Chapter One: Introduction

1 Introduction 1

Chapter Two: Theoretical and Contextual Background

2.0 Introduction 8 2.1 Linguistic variation 8 2.1.1 Place, community and identity 8 2.1.2 Gender 14 2.1.3 Age 21 2.1.4 The group and the individual 24 2.2 Linguistic change 28 2.2.1 Change within a dialect contact framework 29 2.3 Ideology and evaluation 32 2.3.1 Language ideology 32 2.3.2 Evaluation and salience 37

Chapter Three: The Design of the Teesside Study

3.0 Introduction 41 3.1 Sampling decisions 42 3.1.1 Location 43 3.1.1.1 Lack of previous studies 44 3.1.1.2 History of Middlesbrough 45 3.1.1.3 The geographical position of Middlesbrough 51 3.1.1.4 The shifting identity of Middlesbrough 53 3.1.1.5 The current economic and social climate 58 3.1.1.6 The investigation of the researcher’s native variety 60 3.1.2 Dimensions of variation under investigation 60 3.1.3 Design of the fieldwork sample 61 3.2 Method of data elicitation 66 3.2.1 Background to the design of the data elicitation technique 67 3.2.1.1 The proposed SuRE project: aims and difficulties 67 3.2.1.2 Previous studies and their applicability 73 3.2.2 The new method: the SuRE core 76 3.2.2.1 Overall aims 76 3.2.2.2 Sense Relation Network sheets: visual design and content design 78 3.2.2.3 Sense Relation Network sheets: technique of administration 83 3.2.2.4 Sense Relation Network sheets: trialling and revision 85 3.2.2.5 Sense Relation Network sheets: data yielded 86 3.2.2.6 Identity Questionnaire 88 3.2.3 The new method: additions for the Teesside study 91 3.2.3.1 Language Questionnaire 91 3.2.3.2 Teesside Identity Questionnaire 92 3.2.3.3 Affiliation Score Index 93 3.3 Fieldwork 94

Chapter Four: Results: Linguistic Data

4.0 Introduction 98 4.1 Pre-pausal and turn-final (t) 100 4.1.1 Variants under consideration 100 4.1.2 Geographical and social distribution of variants under consideration 101 4.1.3 Particulars of the data and areas of interest 104 4.1.3.1 Some discourse considerations 105

v 4.1.4 Overall distribution of variants of (t): conversational style 107 4.1.5 Pre-pausal (t), conversational style 109 4.1.5.1 Generational variation 109 4.1.5.2 Gender 110 4.1.5.3 Age and gender 111 4.1.6 Turn-final (t), conversational style 115 4.1.6.1 Generational variation 115 4.1.6.2 Gender 117 4.1.6.3 Age and gender 118 4.1.7 Pre-pausal (t): citation forms 112 4.1.8 More discourse considerations 124 4.1.9 Pre-pausal and turn-final (t): summary 127 (p t k) Glottalling and glottalisation 131 4.2.1 Variants under consideration 131 4.2.2 Geographical and social distribution of variants under consideration 132 4.2.3 Particulars of the data and areas of interest 134 4.2.4 Overall distribution of variants of (p), (t) and (k) 135 4.2.5 (p), conversational style 137 4.2.5.1 Generational variation 137 4.2.5.2 Gender 138 4.2.5.3 Age and gender 139 4.2.5.4 (p): citation forms 141 4.2.6 (t), conversational style 143 4.2.6.1 Generational variation 143 4.2.6.2 Gender 144 4.2.6.3 Age and gender 145 4.2.6.4 (t): citation forms 148 4.2.7 (k), conversational style 150 4.2.7.1 Generational variation 150 4.2.7.2 Gender 151 4.2.7.3 Age and gender 153 4.2.7.4 (k): citation forms 155 4.2.8 (p t k): summary 157 4.3 (r) 160 4.3.1 Variants of British English (r) 160 4.3.2 Geographical and social distribution of variants under consideration 161 4.3.3 Particulars of the data and areas of interest 164 4.3.4 Overall distribution of variants of (r) 165 4.3.5 (r), conversational style 166 4.3.5.1 Generational variation 166 4.3.5.2 Gender 167 4.3.5.3 Age and gender 168 4.3.6 (r): citation forms 170 4.3.7 (r): summary 172 4.4 (th)and(dh) 174 4.4.1 Variants of British English (th) and (dh) 174 4.4.2 Geographical and social distribution of variants under consideration 176 4.4.3 Particulars of the data and areas of interest 179 4.4.4 Overall distribution of variants of (th) and (dh) 181 4.4.5 (th), conversational style 181 4.4.5.1 Generational variation 181 4.4.5.2 Age and gender 182 4.4.6 (dh), conversational style 183 4.4.6.1 Generational variation 183 4.4.6.2 Age and gender 183 4.4.7 (th) and (dh): citation forms 184 4.4.8 (th) and (dh): summary 1 86 4.5 Summary of trends 188

Chapter Five: Results: Perceptions of Language and Place

5.0 Introduction 191 5.1 Language 5.1.1 Defining and delimiting MbE 192 5.1.1.1 Defining the accent 192 5.1.1.2 Identifying the accent 193 5.1.1.3 Delimiting the accent 195 5.1.2 Perceptions of variation in MbE 196 5.1.2.1 Perceived age variation 197 5.1.2.2 Perceived gender variation 199 5.1.3 Attitudes towards MbE 201 5.1.3.1 Evaluation of the accent 202 5.1.3.2 Perceived linguistic accommodation 203 5.1.3.3 Perceived misidentification of the accent 205 5.1.4 Summary of perceptions of language 208 5.2 Place 210 5.2.1 Defining and delimiting place 210 5.2.2 Attitudes towards place 214 5.2.3 Perceptions of image 217 5.2.4 Summary of perceptions of place 219 5.3 Strength of local allegiance 220 5.3.1 Summary of strength of local allegiance 223

Chapter Six: Synthesis and Discussion

6.0 Introduction 225 6.1 Ideology 225 6.1.1 Local orientation and linguistic variation 226 6.1.2 Evaluation of variants and linguistic variation 235 6.2 Linguistic variation as a stylistic resource 239 6.2.1 Age 240 6.2.2 Gender 244 6.2.3 Innovatory speaker groups 247 6.3 Summary 249

viii Chapter Seven: Conclusions

7 Conclusions 251

Appendix One: Biographical Data sheet 258 Appendix Two: Instruction sheet 259 Appendix Three: Three completed SRNs 260 Appendix Four: Identity Questionnaire 263 Appendix Five: Completed Language Questionnaire 264 Appendix Six: Affiliation Score Index 265 Appendix Seven: Word List 266 Appendix Eight: Localness charts 267

Appendix Nine: Variants of (th) and (dh) other than [0 ], [f] and [6], [v] 271

References 272

ix List of Figures

3.1 Geographical location of Middlesbrough 43 3.2 Census figures 1801 -1901 47 3.3 Census figures 1901 -1991 51 3.4 Location of Middlesbrough in the North of England 52 3.5 Middlesbrough in the North Riding of Yorkshire 53 3.6 Middlesbrough in Teesside 54 3.7 County Cleveland 55 3.8 Four Local Authorities 56 3.9 U-curve 62 3.10 Variant being introduced 62 3.11 Variant being lost 63 3.12 Sense Relation Network sheet (first of three) 79 3.13 Sense Relation Network sheet (second of three) 80 3.14 Sense Relation Network sheet (third of three) 81 3.15 Identity Questionnaire 89 4.1 Pre-pausal (t), generational differences 109 4.2 Pre-pausal (t), gender 110 4.3 Pre-pausal (t), male speakers 111 4.4 Pre-pausal (t), female speakers 113 4.5 Turn-final (t), generational differences 116 4.6 Turn-final (t), gender 117 4.7 Turn-final (t), male speakers 118 4.8 Turn-final (t), female speakers 120 4.9 (p), generational differences 137 4.10 (p), gender 138 4.11 (p), male speakers 139 4.12 (p), female speakers 140

4.13 (t), generational differences 143 4.14 (t), gender 144

4.15 (t), male speakers 145

x 4.16 (t), female speakers 146 4.17 (k), generational differences 150 4.18 (k), gender 152 4.19 (k), male speakers 153 4.20 (k), female speakers 154 4.21 (r), generational differences 166 4.22 (r), gender 167 4.23 (r), male speakers 168 4.24 (r), female speakers 169 4.25 (th), age and gender (young speakers) 182 4.26 (dh), age and gender (young speakers) 184 5.1 Locations where accent is perceived to be different 195

XI List of Tables

3.1 Design of fieldwork sample 66 4.1 Overall use of variants of (t) in pre-pausal position (conversational style, all speakers) 108 4.2 Overall distribution of variants of (t) in turn-final position (conversational style, all speakers) 108 4.3 Overall distribution of variants of pre-pausal (t): citation forms 122 4.4 Distribution of variants of pre-pausal (t) by age and gender: word list style (WLS) and conversational style (CS) 123 4.5 Number of question tags and sentence tags recorded by age and gender 126 4.6 Percentage use of released, glottalised and glottalled variants of (p t k) (conversational style, all speakers) 135 4.7 Overall distribution of variants of (p), word-list style 141 4.8 Distribution of variants of (p), word-list style (WLS) and conversational style (CS) 142 4.9 Overall distribution of variants of (t), word-list style 148 4.10 Distribution of variants of (t), word-list style (WLS) and conversational style (CS) 149 4.11 Overall distribution of variants of (k), word-list style 156 4.12 Distribution of variants of (t), word-list style (WLS) and conversational style (CS) 156 4.13 Overall distribution of variants of (r) (conversational style, all speakers) 165 4.14 Overall distribution of variants of (r), word-list style 171 4.15 Distribution of variants of (r), (WLS) and (CS) 171 4.16 Overall distribution of variants of (th) and (dh) (conversational style, all speakers) 181 4.17 Distribution of variants of (th) across age 181 4.18 Distribution of variants of (dh) across age 183 4.19 Overall distribution of variants of (th) and (dh) (word-list style) 184 4.20 Distribution of [f] and [v] (word-list style (WLS) and conversational style (CS)) 185

xii 5.1 Definition of accent across age 192 5.2 Informants’ evaluation of accent (male/female) 202 5.3 Informants’ evaluation of accent (age differences) 203 5.4 Perceived advantages of life in Middlesbrough 215 5.5 Perceived drawbacks of life in Middlesbrough 215 5.6 Levels of local allegiance for individual speakers as revealed through responses to IdQ 221 5.7 Mean group scores for Affiliation Score Index 222 Chapter One Introduction

The aims of this study are broadly two-fold. The first aim is methodological. More specifically the objective is to create an innovative method of data elicitation designed for use in studies of social and regional language variation. The utilisation and thus systematic trialling of the method of data elicitation fulfils the second aim of the study: the empirical investigation of a hitherto unresearched urban variety of British English. The variety in question is that of Teesside, in the North East of England.

The methodological setting which underlies the present study of Teesside English has something of a dual purpose. Methodologically, the task at hand was to devise a completely new data elicitation technique which was to become the core of a national collaborative venture1. Although devised with the requirements of a large-scale project in mind, the method has also been developed as an original approach for use in the present research.

The basic intention of the collaborative project was to create a large, computer- held database of consistently-collected dialectal material which would record and document the facts of linguistic variation throughout Britain, permitting detailed analyses of issues concerning the process of language change. The method of data elicitation for the project was to capture data which are analysable on three levels of linguistic variation: phonological, grammatical and lexical. These data were to be comparable across the network of localities to be studied. Furthermore, the technique was to be rapid and uncomplicated to administer, allowing for its use with a socially-stratified sample of speakers from each locality.

1 The present research is undertaken through a White Rose Scholarship, the brief of which was to create a new method of data elicitation for the Survey of Regional English, to be discussed in Chapter 3. 2

The present study of Teesside English, although undertaken within the Labovian paradigm of variationist research, also takes an ethnographic approach to sociolinguistic enquiry. Such an approach can reveal ‘the local knowledge that motivates and explains the behavior of a particular group’ (Johnstone 1999: 19). Rather than a traditional correlational analysis in which linguistic variation is correlated with social groups of informants, speakers’ perceptions of variation and attitudinal information are used to gain insight into the motivation for linguistic variation and the change it implies. To permit qualitative analyses of the data, the method was designed concurrently to elicit attitudinal information for the Teesside study. In terms of the methodological aim of the thesis, two research questions are posed:

1. How best can comparable data on lexical variation be included in a data elicitation technique designed to elicit casual speech in a relatively rapid manner? 2. How best can attitudinal information also be included to allow for qualitative and quantitative analyses of the data?

The context in which the study is set is one of unprecendentedly rapid changes in current British English (henceforth BrE). These changes are argued to be leading to homogenisation on a broad scale (Trudgill 1999a, Foulkes and Docherty in press). The increase in the extent of geographical and social mobility experienced by the population in recent years is frequently cited as a contributory factor in this large-scale homogenisation. The changes, therefore, appear to be best modelled by a levelling process, defined as ‘the eradication of socially or locally marked variants’ (Watt and Milroy 1999: 26), within a dialect contact framework. Thus the changes are seen largely as speaker-based rather than system-based (see Watt 1998, however, for an analysis which integrates internal and external factors influencing vowel variation and changes in Tyneside English). In terms of vowels, this levelling process appears to lead to the development of what can be termed ‘regional standards’ or regional levelled accents (see further Watt 1998, Watt and Milroy 1999, Torgersen and Kerswill 2001). 3

However, several widespread consonantal changes have been identified throughout current BrE, most of which are argued to have a historical association with non-standard varieties of London and the South East of England (Wells 1982, Tollfree 1999, Foulkes and Docherty 1999a). Furthermore, the changes are argued to affect varieties closer to London more than those further removed from the South East of England (Foulkes and Docherty in press). Thus, in terms of consonantal variables at least, evidence to date would appear to support Wells’ claim that London’s ‘working-class accent is today the most influential source of phonological innovation in England and perhaps in the whole English-speaking world’ (1982: 301).

The focus of the present study pertains to variation in consonantal variables suggestive of change in progress in Teesside English. Whether the urban variety of Teesside, situated some 236 miles (378 kilometres) north of London, has been affected by current widespread consonantal changes of BrE is central to the study. Furthermore, the study investigates whether adoption of such forms can, indeed, be said to lead to large-scale homogenisation, or whether evidence can be uncovered of the adoption or increased use of variants which are not associated with the South East of England and which do not form part of the current widespread consonantal changes spreading throughout BrE. Additionally, as ‘the study of meaning in sociolinguistic variation is a study of the relation between variation and identity’ (Eckert 2000: 42), the present research explores whether adoption of forms, particularly ‘Southern’ forms, compromises or affects the speaker’s sense of regional/social identity. Such a question necessitates enquiry into what this sense of identity is and how it is constructed. A locality in which identity construction is complex and fluid affords fertile ground for investigating its connection with language variation. The industrial conurbation around the River Tees, as well as yielding a hitherto unresearched urban variety of BrE, offers an ideal test-site for such a study in language variation and identity.

In a recent article in a British national newspaper, Middlesbrough, the most densely populated urban centre on Teesside and the locality for the present study, was described thus: 4

[i]t is pretty much a place between places. It’s not on the way to anywhere, it’s not quite in Yorkshire, and in fact a lot of people don’t know where it is. It is a forgotten part of Britain with no identity. (The Sunday Times, March 5, 2000).

Situated some 38 miles (61 kms) south of Newcastle in the North East of England and around 50 miles (80 kms) north of York in Yorkshire, Middlesbrough lies in something of a transition area between the lower part of the North East of England and the upper part of Yorkshire. The transitional nature of the geographical location of Middlesbrough, with the sense of it being neither wholly in one region nor the other, has meant that the identity of the urban centre is not deep-rooted, or firmly felt by outsiders. In fact, the precise identity of Middlesbrough evokes a degree of uncertainty and confusion even amongst its inhabitants.

A large part of this uncertainty can be traced to the redrawing of local administrative boundaries in the region. Traditionally, the River Tees divided the conurbation, with urban centres lying north of the river being situated in County Durham and those on the south bank, including Middlesbrough, forming part of the North Riding of Yorkshire. The conurbation was brought together in 1968 with the formation of Teesside, expanded in 1974 with the formation of Cleveland and divided once more in 1996 with the formation of four separate Local Authorities. The urban centres of the conurbation have thus been assigned four separate identities in the past 33 years. Coupled with the changes in identity in terms of local administrative boundaries, the area has witnessed a shift in identity in terms of popular culture as the conurbation, both north and south of the Tees, has become increasingly associated with the North East of England and increasingly dissociated from Yorkshire. The effect of the fluid nature of the identity of the region on the inhabitants’ self-images, and any modification of linguistic behaviour in line with the changing self-images, are therefore central to the study. Moreover, combined with the problematic construction of identity in the urban centre in question, in the speech of the young on Teesside a tension is predicted between constructing and retaining a local identity and appearing outward-looking through the use of variants which form part of the current changes identified in BrE. 5

In terms of the second aim of the study then, the empirical investigation of variation in Teesside English, the research questions posed are:

3. Are innovatory forms in evidence in the variety in question? 4. What is the nature of socially-conditioned variation in the consonantal variables of the study? 5. To what extent do the socially-conditioned linguistic patterns correlate with speakers’ identity construction and local orientation?

Before looking at the particulars of the Teesside study, aspects of the theoretical and contextual background in which the study is grounded are detailed in Chapter 2. The parameters of variation pertinent to the study are examined. The extra-linguistic variable of gender is discussed in terms of the approaches developed to model gender differences in linguistic behaviour. It is argued that viewing gender as a dichotomised opposition, polarising gender and asserting that males speak one way and females speak another, can lead to over­ simplification and ignore the heterogeneity of genderlects. Similarly, age as an extra-linguistic variable is considered, with emphasis placed on the necessity of viewing age in terms of life stages rather than simply chronological numbers (Eckert 1998). Additionally, the conceptualisation of place and the notion of community are explored, with the relativity of the concept of place and the significance of perceived and social space discussed. Aspects of speaker-based language change within a dialect contact framework are also outlined, and the processes of levelling and diffusion are considered. Furthermore, the concept of ideology and its application to linguistic investigation are summarised.

The innovative method of data elicitation designed for use in the collaborative Survey of Regional English project is described in Chapter 3. Included is the design and technique of administration of the Sense Relation Network sheets and the Identity Questionnaire which form the core of the interview. Additional elements designed and included specifically for the Teesside study are also outlined. The sampling decisions of the study are detailed, with justification for the choice of locality. This includes the historical development of the urban centre, its geographical position, and, importantly, its shifting identity both in 6 terms of local administration boundaries and in terms of popular culture. The design of the fieldwork sample is also outlined, detailing the emically-defined age cohorts used in the study.

We turn to the findings of the study in Chapter 4 in which variation in evidence in the linguistic data is detailed. The variation is considered in terms o f the overall distribution of variants under investigation. How the extra-linguistic variables of age and gender act on the data is then analysed. Generational differences which are suggestive of change in progress in the variety are presented, as are fine-grained age differentiations in the data. Evidence of style shifting is also sought, as the distribution of variants in the more formal word-list style are compared with the patterning revealed in conversational style.

Insight into the motivation for the language variation presented in Chapter 4 is gained through attempting to access the local knowledge that speakers operate with when constructing and projecting their sociolinguistic identities. Chapter 5 deals with the attitudinal findings of the study. The speakers’ perceptions of language and place provide insight into the construction of community and the collective identity of the informants as well as age- and gender-correlated variation within this collective identity construction. How the speech community is defined and delimited is analysed, as is awareness of linguistic variation within the community. Local orientations and levels of local allegiance are also explored.

A synthesis of the linguistic data and the attitudinal data is presented in Chapter 6. A language ideology framework, based on linguistic anthropological theories, is invoked to account for the variation and suggested change in Middlesbrough English (henceforth MbE). Assessment of speakers’ opinions and overt discussion and awareness of language variation allows insight into the symbolic function of the dialect. The observations about language and community suggest that forms are reallocated to index locally relevant groups. The local orientation of the speakers is considered in terms of the adoption and increasing use of linguistic forms. Additionally, how variants are evaluated is examined through overt comment and evidence of style shifting. The use of linguistic variation as a 7

stylistic resource is analysed and speakers’ perceptions of age- and gender- correlated differences in language are viewed in the light of actual socially- conditioned variation in evidence in the linguistic data. Finally Chapter 7 assesses the extent to which the aims of the study have been fulfilled and considers conclusions that can be drawn from the results. Chapter Two Theoretical and Contextual Background

2.0 Introduction

In this chapter the theoretical and contextual background against which the study is set will be outlined. The parameters of variation of relevance to the present study will be discussed in section 2.1. Additionally, processes of and motivation for speaker-based language change will be detailed in section 2.2. Finally, aspects of language ideology which will form the interpretative framework for the findings are outlined in section 2.3.

2.1 Linguistic variation

The list of external factors which may affect linguistic behaviour could doubtless be extended indefinitely. Therefore the parameters used to divide the sample population under investigation must be chosen in a principled way. Reference in this section is limited to the external factors of interest to the study. The independent social variables utilised are gender and age, which are considered in sections 2.1.2 and 2.1.3 respectively. However, viewing speakers as individuals as well as parts of a group is crucial, as ‘[cjlass, sex, age, region, the nature of the linguistic task, and the makeup of the audience all have an important bearing on how people sound; but they do not d e t e r m in e how people sound’ (Johnstone and Bean 1997: 236). Therefore, the role of the individual speaker within the group is considered in section 2.1.4. The notions of place, speech community and identity play a central part in the study, however, and we begin with an examination of such concepts.

2.1.1 Place, community and identity

In diachronic or traditional dialectological studies of language variation, in methodological terms the extra-linguistic variable of interest was geographical 9 place. In studies such as the Survey of English Dialects (1962-71) (henceforth SED), the geographical distribution of variables was studied. Social variation within a locality was not explicitly investigated. As 55% of the questions in the SED interview were lexical and 29% were phonological (Orton 1962: 15), the principal interest was the geographical distribution of lexical and phonological forms. One manner in which the data were presented was in the form of linguistic maps (for example, Orton, Sanderson & Widdowson 1978). Research within the traditional dialectological paradigm was historical in orientation. Fieldwork was carried out mainly in rural localities, the aim being the recording of conservative, non-standard forms of the language. The sampling decisions reflected this aim in predominantly targeting non-mobile, older, rural, males (NORMs, see Chambers and Trudgill 1980: 33-5) who, it was felt, were the speakers most likely to use ‘vernacular’ consistently and genuinely (Orton 1962: 15). (In the event, not all the informants used in the survey fulfilled such strict criteria, see Coates 1993, Johnston 1985a and Stoddart, Upton & Widdowson 1999 for further details.) The interest in geographical variation led to much criticism of the methods employed by traditional dialectological work such as the SED and, as Chambers & Trudgill (1980: 35) argue, ‘the narrow choice of informants in dialect geography is probably...the greatest single source of disaffection for it in recent times’.

Although it is now widely acknowledged that dialects are both regional and social, ‘since all speakers have a social background as well as a regional location’ (Chambers & Trudgill 1980: 54), it was not until Labov’s seminal works on Martha’s Vineyard (1963) and New York City (1966) that social variation within a single geographical location was investigated in detail. Work undertaken within a Labovian paradigm has since been dominant in the investigation of language variation and change.

Although the shift has been away from researching purely geographical variation in recent years, in the investigation of one geographical locality the notion of place is still relevant. Much debate has centred on defining the notion of the speech community. It is not within the scope of this work to detail all definitions and criticisms of the notion. However, a broad outline will be given of the notion 10

(the reader is referred to Gumperz 1968, Romaine 1982, Dorian 1982, Wardhaugh 1992, Hudson 1980 for further discussion).

Gumperz first defined the speech community as a sociolinguistic entity:

[i]n analyzing linguistic phenomena within a socially defined uni verse... the study is of language usage as it reflects more general behavior norms. This universe is the speech community: any human aggregate characterized by regular and frequent interaction by means of a shared body of verbal signs and set off from similar aggregates by significant differences in language usage (Gumperz 1968: 381).

Since this definition, references to the speech community in sociolinguistic literature have been rife. The notion is not unproblematic, however. Exactly how a speech community is defined and delimited is unclear. Rather than initial definitions of speech communities which were based solely on their linguistic characteristics, emphasis was later placed on the speech community as a group of people who perceive themselves to be a community, for example, Labov’s (1972: 120-1) definition:

[t]he speech community is not defined by any marked agreement in the use of language elements, so much as by participation in a set of shared norms; these norms may be observed in overt types of evaluative behaviour, and by the uniformity of abstract patterns of variation which are invariant in respect to particular levels of usage.

The set of shared norms, however, may vary considerably among sub-groups of the community. Thus, a degree of ‘fuzziness’ as to the precise characteristics of the concept of a ‘speech community’ can be seen to exist.

If we take the view that speech communities should have some kind of psychological reality for their members, then the individual speaker should perceive such communities to exist in society. Furthermore, the concepts of ‘group’ or ‘community’ are relative in the sense that an individual belongs to various speech communities at the same time. Depending on the context, 11 individuals may identify with any one of the various speech communities they feel themselves to belong to. Brown and Levinson (1979: 298-9) argue:

[sjocial theorists who adopt the stronger concept of structure are more likely to think of groups as relative concepts, each group being a unit that is relevant only in relation to units of like size that for immediate purposes are contrasted with it. Thus for a man who lives in Cambridge, his territorial identification will be with Cambridge when contrasted with Newmarket, with Cambridgeshire when contrasted with Lancashire, with England when contrasted with Scotland, with the United Kingdom when contrasted with Germany, and so on.

Recently, Irvine and Gal (2000: 75) have argued that the concept of a ‘speech community’, although useful for understanding the organisation of local repertoires, neglects larger boundary relationships, cultural oppositions, borders and conflict. According to Irvine and Gal, the ways in which identity is produced by ideas of opposition between culturally defined groups and by practices that promote divergence and differentiation are rarely examined.

Furthermore, which groups the speaker identifies with is clearly crucial to an understanding of what ‘speech community’ has psychological reality for the speaker. Le Page and Tabouret-Keller’s theory of acts of identity is a useful concept in this regard. Le Page and Tabouret-Keller consider ‘linguistic behaviour as a series of acts of identity in which people reveal both their personal identity and their search for social roles’ (1985: 14). Based on a study of the language of children in multilingual Belize, Le Page and Tabouret-Keller intended the theory of acts of identity to be ‘universally applicable’ (1985: 182). Le Page and Tabouret-Keller’s (1985: 181) modified hypothesis states that:

the individual creates for himself the patterns of his linguistic behaviour so as to resemble those of the group or groups with which from time to time he wishes to be identified, or so as to be unlike those from whom he wishes to be distinguished.

Further, constraints upon speakers’ acts of identity are identified: 12

we can only behave according to the behavioural patterns of groups we find it desirable to identify with to the extent that: (i) we can identify the groups (ii) we have both adequate access to the groups and ability to analyse their behavioural patterns (iii) the motivation to join the groups is sufficiently powerful, and is either reinforced or reversed by feedback from the groups (iv) we have the ability to modify our behaviour.

Hence the concept of the ‘speech community’, in terms of which communities or groups speakers identify with or perceive themselves to belong to, is not immediately accessible to the researcher. The construction of community in terms of Self and Other, the relativity of the concept of speech community, and the concept that a speaker’s choice of variant represents an act of identity with groups with whom s/he wishes to be identified are of importance to the present research and will be considered further in Chapter 6.

The concept of place is likewise not unproblematic. In traditional geography, place is seen as physical and identifiable by a set of co-ordinates on a map. Regions can also be delimited on the basis of economic or historical criteria. However, more recently how place is conceptualised has been re-evaluated. What makes regions real is a question which has come to concern humanistic geographers. In most dialectological and variationist research, place has been viewed in objective, physical terms. More recently, in ethnographic studies of variation, how place is socially constructed has been investigated (see for example Eckert 2000). Identifying where someone is from is not as objective and categorical a matter as it may seem. Rather, accessing the psychological reality of where an individual can be said to be from is considerably more complex. According to Johnstone (1999), regions are voluntary, as opposed to traditional, in that they are the result of how humans choose to divide up the world they experience. Thus ‘speakers are seen as constructing place as they experience physical and social space, and different speakers may orient to place, linguistically, in very different ways for very different purposes’ (Johnstone 1999: 2). Preston’s (1989) folk dialectological work uses mapping tasks to investigate how people construct and relate to different meaningful regions. In BrE, the identification of dialects and speech community focussing has recently 13 been investigated by Kerswill & Williams (2000), which we consider further in section 2.3.2.

Britain (2000: 38) argues that in the past space has been regarded as ‘a blank stage on which sociolinguistic processes are enacted’. In a further refinement to the understanding of geographical space, he distinguishes three types of space:

Euclidean space - the objective, geometric, socially divorced space of mathematics and physics.

Social space - this is the space shaped by social organisation and human agency, by the human manipulation of the landscape, by the contextualisation of face-to-face interaction, by the creation of a built environment, and by the relationship of these to the way the state spatially organises and controls at a political level.

Perceived space: this is how civil society perceives its immediate and not so immediate environments - important given the way people’s environmental perceptions and attitudes construct and are constructed by everyday practice (Britain 2000: 39-40).

Britain’s work on the Fens shows that, although the Euclidean distance to London is similar, the neighbouring city of Peterborough is more affected by various changes diffusing from London than are the Fens. As well as the affect of geographical distance, contact between inhabitants of the Fens and Peterborough is restricted by poor transport facilities and negative reactions (the perception of inhabitants of the Fens as ‘carrot crunchers’ (2000: 57), for example), as well as geographical distance. Britain argues that it is not just the traditional concept of Euclidean space which influences the maintenance or change of linguistic forms, but social space and perceived space also affect linguistic behaviour. Again, the concept of place in terms of perceived space and social space is of particular relevance to work on Teesside and will be considered in more detail in Chapter 14

2.1.2 Gender

Differences between male and female language have long been acknowledged by linguists. Early discussion on such variation drew heavily on folklinguistic beliefs of how males and females use language differently, with little scientific evidence provided to support the beliefs held, for example, Jesperson’s assertion that women talk more than men and demonstrate an ‘instinctive shrinking from coarse and gross expressions’ (1922: 246). Since the publication of Lakoffs (1975) ‘Language and Woman’s Place’, interest in the difference between how males and females use language has not diminished, and debate among researchers on the adequacy of theories which seek to explain why or if males and females use language differently continues.

Research on male/female differences in speech has largely focussed on two domains of language behaviour: variation at the phonological level and variation in conversational styles. Three distinct approaches have been developed to model the differences between male and female speech; the deficit model, the power/dominance approach and the culture/difference approach. With its assertion that female speech is weak and unassertive, Lakoff s work is an example of research illustrating the deficit model. Through this approach, female speech patterns are seen as inferior when compared with male discourse, which is considered the norm. The work of Zimmerman and West (1975) and Spender (1980) exemplifies the power/dominance approach. Female speech is not viewed as inferior to or deviant from male speech through this approach. Rather, the male/female difference is seen as a product of the asymmetrical power relations in society. The economic and social power held by males in society is thought to permeate into language, and the sexual division of labour is thought not only to cause the male/female difference in speech, but also to maintain it. The culture/difference approach evolved with the work of Maltz and Borker (1982), who developed a theory of miscommunication based on the work of Gumperz (1982). Maltz and Borker claimed that male/female differences in speech patterns are a consequence of cultural difference which is implemented during the socialisation process. Miscommunication thus occurs in mixed-sex adult interaction. 15

All three approaches have been criticised heavily. It is not within the scope of this work to explore the various approaches and their criticisms in detail (see Bergvall, Bing and Freed 1996, Bergvall 1999 for further discussion). However, the view that the complexity of language and gender is considerably oversimplified by viewing males and females in a dichotomised manner is acceded (see further Cameron 1995, Freed 1996, Mullany 2000). The concentration on differences between male and female speech leads to a disregard of the homogeneity between the speech patterns of the two genders. Furthermore, polarising gender and asserting that males speak one way and females speak another, as the three approaches outlined tend to do, ignores not only the heterogeneity within male speech or female speech but also the interaction between gender and other social variables. This appears particularly relevant to studies of variation at the phonological level. Indeed, Eckert (2000), in her study of the vowels of Detroit teenagers, in some cases found greater differences between groups of females than when comparing female speech with male speech. She argues ‘the primary importance of gender lies not in differences between male and female across the board, but in differences within gender groups’ (2000: 122-123). The interaction of social variables resulting in the heterogeneity of linguistic behaviour within a single-sex group is of importance to the present research, and we will return to it within the course of the work.

Viewing gender as a performative social construct is one way of avoiding the polarisation of gender (Butler 1990). The notion of performativity allows for an amount of flexibility in the consideration of gender, rather than viewing gender as a fixed and rigid given. Rather than learning and then mechanically reproducing ways of speaking appropriate to their own sex, males and females learn a complex set of gendered meanings and are capable of using features associated with either masculinity or femininity (see Cameron 1997, Bergvall, Bing and Freed 1996 for further discussion).

Criticism also of the analysis of variation based on sex rather than gender has been made (see particularly Eckert 1989). Sex is defined as ‘biological or anatomical differences between men and women’ whereas gender ‘concerns the 16 psychological, social and cultural differences between males and females’ (Giddens 1989: 128). When analysing differences in linguistic behaviour we are largely concerned with gender variation. Little evidence exists for physiologically- or neurologically-based sex differences in language (for overviews see Chambers 1995: 132-137, Watt 1998: 89-90). One of the few arguments put forward for sex-based variation can be found in Chambers, who reviews data from psychological literature which reveal females to have an advantage over males in a variety of tests of linguistic abilities. As similar male/female differences can be found in Middle Class (MC) industrial societies where there is a blurring of gender roles as in Working Class (WC) societies where traditional gender roles persist, Chambers maintains:

[i]n these instances, it is plausible to speculate that we are seeing another effect of the sex-based difference: the neuropsychological verbal advantage of females results in sociolinguistic discrepancies such that women use a larger repertoire of variants and command a wider range of styles than men of the same social groups even though gender roles are similar or identical. The tendency for women to take on gender roles requiring greater mobility may be a result of their innate sociolinguistic advantage rather than a cause of it [sic] (Chambers 1995: 136-137).

This view is criticised by Labov, however, who concludes, ‘[i]t seems hard to avoid the conclusion that women’s linguistic conformity is primarily a social, not a biological, phenomenon’ (2001: 277). Despite the relatively sparse evidence for sex-based variation in speech and the fact that gender variation is largely what is of interest in variationist studies, the binary category of biological sex is that which is commonly used to analyse and explain gender variation. Eckert maintains:

although an individual’s gender-related place in society is a multidimensional complex that can only be characterised through careful analysis, his or her sex is generally a readily observable binary variable, and inasmuch as sex can be said to be a rough statistical indication of gender, it has been reasonable to substitute the biological category for the social in sampling. However, because information about the individual’s sex is easily accessible, data can be gathered without any inquiry into the construction of gender in that community. As a result, since 17

researchers have not had to struggle to find the categories in question, they tend to fall back on unanalyzed notions about gender to interpret whatever sex correlations emerge in the data and not to consider gender when there are no sex correlations (Eckert 1989: 247).

Within variationist studies of speech communities the significance of gender- correlated variability has been repeatedly shown, with numerous studies revealing the different exploitation of resources by male and female speakers (see for example Labov 1966, 1990, Trudgill 1974, Macaulay 1978, Cheshire 1982, Milroy 1987a, Docherty et al. 1997, Kerswill and Williams 1997, Williams and Kerswill 1999). Perhaps the most frequently attested observation about language variation with regard to gender is the tendency of females to avoid stigmatised forms. Chambers (1996: 102) states that:

[i]n virtually all sociolinguistic studies that include a sample of males and females, there is evidence for this conclusion about their linguistic behaviour: women use fewer stigmatised and non­ standard variants than do men of the same social group in the same circumstances.

Indeed, this was found to be the case in many early studies. Trudgill’s (1974) findings from Norwich followed this pattern and were confirmed by a self- evaluation test in which speakers were asked whether they used standard or non­ standard forms of certain linguistic features. It was found that male speakers over-reported their use of non-standard forms whilst females over-reported their use of standard forms. This is explained by Trudgill in terms of different prestiges: ‘[because society evaluates different characteristics differently in the two sexes, covert prestige exerts a more powerful influence on men, and ‘normal’ prestige on women.’ (2000: 77). Why this should be so is explained by Trudgill in terms of females’ powerless position in society and the connotations of masculinity associated with WC culture. Trudgill (1997: 183) argues:

[t]he social position of women in our society is less securc than that of men, and, generally speaking, subordinate to men. It is therefore more necessary for women to secure and signal their social status linguistically in other ways. 18

On the other hand, ‘roughness and toughness’ which are seen as ‘desirable masculine attributes’ and are associated with WC life, and thus with WC speech, are not considered desirable in females, as ‘refinement and sophistication are much preferred’ (1997: 183). Such explanations have been criticised by various researchers, assuming as they do that gender can be subsumed under the variable of class, and that the prestige language is male language (see further Coates 1993, Spender 1980, Cameron 1990). For example, an obvious question is begged in the apparent contradiction of why, in signalling their social status linguistically, women, given their subordinate position to men, do not imitate the speech patterns of males and use low prestige forms.

Labov (2001: 266-293) discusses three principles concerning the variation between male and female linguistic behaviour. These are:

Principle 2, the linguistic conformity of women: For stable sociolinguistic variables, women show a lower rate of stigmatised variants and a higher rate of prestige variants than men.

Principle 3 In linguistic change from above, women adopt prestige forms at a higher rate than men.

Principle 4 In linguistic change from below, women use higher frequencies of innovative forms than men do.

The juxtaposition of these three principles is seen by Labov as the Gender Paradox: ‘[w]omen conform more closely than men to sociolinguistic norms that are overtly prescribed, but conform less than men when they are not’ (2001: 293).

In early variationist studies, the social variable of class or socio-economic position was seen as primary. Gender variation was considered largely in the context of its interaction with social class. The focus was on the use of vernacular forms and the use of standard forms, and the Lower Middle Class (LMC) females were often found to hypercorrect in their use of standard forms, which was argued to lead to linguistic change. More recently, the significance of 19 the social variable of gender has become increasingly apparent, and Milroy, Milroy and Hartley (1994: 10) argue, ‘there is now a body of evidence suggesting that gender may be prior to class in its capacity to account for socially distributed patterns of language variation Rather than the interaction of gender and social class, the interaction of gender and social category affiliation, and consideration of gender-based differences in the types of social network ties which exist in the community under investigation, have been analysed to effect by researchers.

The interaction of gender and social category affiliation is explored by Eckert (2000) in Detroit. Eckert uses the construct of communities of practice to categorise the adolescents of her study into three social groups (to be considered in more detail in section 2.1.4). The phonological variation investigated by Eckert involved vocalic changes which are part of the Northern Cities Chain Shift in the United States (see section 2.2). Eckert found that the variables which showed the most marked differentiation between genders were the older, less vital changes which are ‘probably not very effective as carriers of counteradult adolescent meaning’ (1997: 225). In the newer changes, for example the backing of (uh), social category differences appeared more marked. In both sets of changes females were found to use a greater range of variants than males. However, group membership appeared more salient for females than males, with girls ‘asserting their category identities through language more than boys’ (Eckert 1997: 226). This, Eckert argues, is ‘the adolescent manifestation of the broader generalization that women, deprived of access to real power, must claim status through the use of symbols of social membership’.

The significance of an analysis of types of social network ties associated with gender in the community under investigation is demonstrated in the work of J. Milroy (1992) and L. Milroy (1987a) in Belfast. The notion of social network will be considered further in section 2.1.4. What is of importance to the present section in the Belfast study is the changing gender roles in the communities under investigation by the Milroys. The socio-economic climate in Clonard, a Catholic community of interest, had changed drastically and considerably more than that in the Protestant community under investigation, Ballymacarrett. 20

Unlike in Ballymacarrett, in the Clonard community massive unemployment had meant that the traditional WC gender roles of men going out to work and women staying at home had changed, and young women had begun to leave home to find work. The transmission of new forms, for example, backed variants of (a), across the sectarian boundary of East to West Belfast was found to be through the Clonard women who, unlike their male counterparts, were in full employment outside the Clonard community. Working in a city-centre store, the Clonard women were likely to come into contact with large numbers of people from all over Belfast, both Catholic and Protestant. Their exposure to innovatory forms was frequent, and thus the transmission of new variants was dependent on the sociolinguistic behaviour of females and the realignment of traditional gender roles.

Recent studies of gender-correlated variation have found the prestige-based vernacular/standard continuum to be somewhat simplistic and less relevant than a local/supra-local opposition which recognises gradations in terms of local and non-local. Milroy, Milroy and Hartley (1994: 2), in their study of glottalisation in Tyneside schoolchildren, hypothesised that ‘males and females will show relative differences in their use of localised and supra-local variants, with males preferring the more local variants’. This was indeed found to be the case, with boys favouring the glottal reinforced variant whilst girls favoured the glottal stop. Similar findings were uncovered in the adult community of Tyneside in terms of glottalisation (see Docherty et al. 1997, Trousdale 2000) and in vowels (Watt 1998, Watt and Milroy 1999). The female speakers of the Tyneside study were found to be greatly in advance of the males in preferring the unmarked northern variants of the f a c e and g o a t 1 vowels (Watt and Milroy 1999: 41). Such a finding was found to be ‘in line with variationist studies which have generally identified women as innovators’ (Watt and Milroy 1999: 41).

In the pilot study of Middlesbrough (Llamas 1998), in which gender was the extra-linguistic variable of interest (a sample of four male and four female WC

' Ihe keywords employed by Wells (1982) to characterise the standard lexical sets of words which share the same vowel are used in the present study. 21 speakers aged between 15 and 30 was used in the study), significant gender differences were found in the majority of linguistic variables under investigation. Gender differentiation in patterns of variation in MbE, both actual and perceived, is therefore a central concern of the present study.

2.1.3 Age

Age is also taken as an independent social variable in this research. The extra- linguistic variable of age has long been used in variationist studies of language. Age-correlated variation can reflect change in progress, change in the speech of the community as it moves through time, and age-grading, change in the speech of the individual as s/he moves through life.

Age-grading is described as ‘the characteristic linguistic behaviours which are appropriate to and typical of different stages in the speaker’s life-span’ (Romaine 1984: 104). Hudson goes further, to connect age-grading with the speech of children: ‘there are linguistic forms which are used only by children in the peer- oriented stage, and which are transmitted from one generation of children to the next without ever being used by adults’ (1980: 16). Chambers, who states that age-graded changes are ‘changes in the use of a variant that recur at a particular age in successive generations’ (1995: 188), claims that very few age-graded changes have been reported.

Age-correlated variation which is suggestive of change in progress in the community can be detected by making a series of observations of similar populations over time. The value of these real-time studies depends on the extent to which the samples observed are comparable. Two kinds of real-time studies can be used. A panel study follows the same individuals and monitors any changes in their behaviour by ‘submitting them to the same questionnaire, interview or experiment’ (Labov 1994: 76). The age cohort is tracked as it moves through time. Thus, a panel study shows change in the individual’s lifetime (cf. Mees & Collins 1999). A trend study tracks the speech of different but comparable individuals, offering a study of a life stage as experienced by successive age cohorts. When used with an age-graded sample, the trend study 22 can provide the most reliable evidence of change in progress (cf. Trudgill 1988). However, practicalities are such that it is rarely possible to undertake a large- scale trend study with a stratified sample of speakers. An apparent time study is more commonly used in variationist research. An apparent time study involves the observation of the speech of different age groups simultaneously. Such apparent-time studies have ‘the inestimable advantage of making information about temporal developments available in a shorter time than the developments themselves take’ (Chambers 1995: 193). However, the inferences made from the results of such studies depend upon the apparent-time hypothesis: ‘the linguistic usage of a certain age group will remain essentially the same for the people in that group as they grow older’ (Chambers 1995: 194). Whether the age- correlated variation revealed in an apparent-time study reflects change in progress or age-grading is not always beyond doubt. Eckert argues that to resolve the ambiguity between age-grading and change in apparent time an understanding of the linguistic life course is needed. This involves consideration of issues such as to what extent a speaker’s language can change over the life course, and how these changes are embedded in life stages and life events (Eckert 1998: 152). Chambers, however, sees the ambiguity as less problematic as ‘[s]ince age-grading is relatively rare and is realised in a distinctive, identifiable pattern, it does not refute the hypothesis but is a codicil on it’ (Chambers 1995: 194).

Of the life stages experienced by speakers, perhaps most research in recent years has concentrated on the speech of children and adolescents. Studies have focussed on the speech patterns of children and adolescents (Romaine 1984, Reid 1978, Macaulay 1978, Cheshire 1982, Roberts and Labov 1995, Eckert 2000, Kerswill & Williams 1994, Foulkes, Docherty & Watt 1999). The interest in the language of children and adolescents has been on the development and acquisition of vernacular forms and structured variation. The speech of adolescents is of particular interest as it is often thought that adolescence is the time in which use of socially stigmatised forms is at its maximum (Romaine 1984). Peer groups are considered to exert great normative pressure at this life stage and this is combined with a lowered susceptibility to society-wide norms as represented by the institutions of the adult and outside world (Downes 1984). 23

Downes goes further, to suggest that this affects males more than females: ‘[i]f one wants to observe the most extreme forms of vernacular speech, the place to look is among male adolescent peer groups’ (1984: 190).

In the search for evidence of potential change in progress, the speech of adolescents is seen as particularly important, as it is widely agreed that adolescence is the focal point of linguistic innovation and change (Chambers 1995: 176). Indeed, Eckert (1998: 163) states:

[ajdolescents lead the entire age spectrum in sound change and in the general use of vernacular variables, and this lead is attributed to adolescents’ engagement in constructing identities in opposition to - or at least independently of - their elders.

In contrast, young adulthood is seen as a time in which the sociolinguistic range takes the form o f‘retrenchment following the adolescent years’ Chambers (1995: 177). Adults, generally, are thought to use more prestige or conservative variants than younger speakers since ‘ [i]f adolescence is the life stage in which speakers push the envelope of variation, conservatism is said to set in during adulthood’ (Eckert 1998: 164). This phenomenon has been revealed in many studies which compare the speech of adults with younger speakers (Labov 1966, Trudgill 1974, Macaulay 1978, Williams and Kerswill 1999, Kerswill and Williams 2000, Docherty et al. 1997). Within the working life stage, the pressures from outside become greater. Hence the use of prestige forms is thought to peak in the middle years, or between 30 and 55, when the maximum societal pressure to conform is thought to be felt (Holmes 1992: 186). These pressures taper off in later years, and the use of prestige forms gives way to a favouring of the vernacular in older speakers. When a sociolinguistic variable is not in the process of change, then, a U-curve is the normal pattern of distribution, in which young and older speakers characteristically use higher proportions of lower-status variants than middle aged speakers (See Figure 3.9 in the following chapter).

Studies commonly rely on chronological age to group speakers. As the span of ages of an individual’s life is so great, speakers are generally grouped in fairly broad cohorts, it being impracticable to design studies which seek fine-grained 24 age differentiations over a range of life stages. This is most notable in the adult groupings, where it is not always clear what the relevant age categories are. Age cohorts can be defined either etically or emically. Etically-defined cohorts are separated into equal age spans, for example decades. Emically-defined cohorts, on the other hand, are grouped according to a shared experience of time. This can be a life stage, for example, adolescence, middle age, or can be related to some aspect of history as ‘[t]o the extent that social and political events can affect the way people speak, age differences in variation can reflect social and political change’ (Eckert 1998: 166). So, life stage and cohort membership can be invoked to explain behaviour, an example of this given by Eckert is the fiscal conservatism of middle aged Americans whose childhoods had coincided with the Depression.

In the design of the pilot study of Middlesbrough (Llamas 1998), age was not taken as an extra-linguistic variable. However, marked age-correlated differences were recorded between two 30 year old speakers and the rest of the sample. As well as generational variation which is suggestive of linguistic change in progress, fine-grained age differentiations which correspond to life stages are of particular interest in the present research. Age will thus be considered further in the design of the study, the results, and the interpretations of the results.

2.1.4 The group and the individual

Although we have considered the notion of ‘group’ indirectly in section 2.1.1, further consideration of the concept of the group and how the individual relates to the group is necessary when considering patterns of linguistic variation. Group, like speech community, is a term which is difficult to define precisely (see Brown and Levinson 1979: 298-9). Although in the present study individuals are grouped by age and gender, it is acknowledged that neither age nor gender can be said to constitute a social group as such. Coates (1993: 7-8) lists the following observations about ‘normal’ groups which have been the object of study in sociolinguistic research, for example, adolescents in Reading (Cheshire 1982), WC residents of Belfast (Milroy 1987a): 25

1. Either they live near each other in a recognised neighbourhood .. or they have a recognised meeting place. 2. They have a recognisable and distinctive sub-culture. 3. Members of the group acknowledge the existence of the group: belonging to the group is part of their identity.

According to Coates, women do not meet these criteria as a social group. Arguably neither do arbitrarily delimited age groups. Nonetheless, the extra- linguistic variables of gender and age are used to group individuals within what can be seen as a speech community.

An arguably more socially meaningful group can be seen in the concept of networks. The sociolinguistic application of the concept of networks was introduced by the Milroys in their study of Belfast (L. Milroy 1987a, J. Milroy 1992). The sociological concept was invoked ‘to explain individual behaviour of various kinds which cannot be accounted for in terms of corporate group membership’ (Milroy 1987a: 135). The concept of network was based not on comparisons between groups of speakers, but on relationships of individual speakers with other individuals. The Belfast study focussed on the maintenance of local norms and the relevance of this to linguistic change in progress. The concept of social network was used both as a quantitative speaker variable and an interpretative category. Network ties were measured in terms of both density, the number of links among individuals, and multiplexity, the content of the links among individuals. Strong network ties were seen as norm-enforcing mechanisms whilst individuals with weak network ties could function as conduits of linguistic change.

More recently, Eckert has used the construct of communities of practice, defined as ‘an aggregate of people who come together around some enterprise’ (2000: 35), to categorise the Detroit adolescents of her study into three social categories: the Jocks, the Burnouts and the In-betweens. Eckert argues that:

[v]iewing speakers in terms of the communities of practice that they participate in recognizes the fluidity and complexity of identity and social participation, pulling us away from a tendency to “pigeon-hole” speakers....it is the collection of types of communities of practice at different places in society that 26

ultimately constitutes the assemblage of practice that is viewed as class, culture, ethnic culture, gender practice, etc. (Eckert 2000: 39).

When analysing linguistic data in groupings of whatever choosing, outliers within a particular group can, on occasion, be identified. These outlying individual speakers can be found to have distinctive characteristics which are seen as being central to a particular role played in the process of linguistic change. However, it is not the intention of an investigation of linguistic variation and change to search for and identify individual speakers. Rather, the search for ‘social locations and social types’ (Labov 2001: 33) is what is of interest, as:

[t]he leaders of linguistic change are not individual inventors of a certain form, but rather those who, by reason of their social histories and patterns of behavior, will advance the ongoing change most strongly (Labov 2001: 33-4).

Various labels have been used to identify characteristics of individuals who are seen to play a particular role in linguistic maintenance and change. Chambers (1995) identifies four distinctive minorities based on sociolinguistic literature: interlopers, insiders, outsiders, and aspirers. Interlopers are individuals who move from one dialect area to another and fail to master highly complex dialect features of the new locality (see Payne’s 1980 work in King of Prussia). Chambers (1995: 85) argues that if the move is made before the age of seven, the new dialect will be mastered as a native speaker. However, if it is made after the age of 14 the speaker’s non-native origins will always be betrayed. In between the ages of seven and 14 mastery of the dialect is variable. Insiders are individuals who are at the centre of their social group but are not necessarily core members or leaders. According to Chambers (1995: 91), insiders ‘embody the social characteristics of their group prototypically, and actualize the linguistic trends in the data prolifically’. Outsiders, as the label might suggest, are individuals apart from the group who occasionally associate with others as fringe members of the group. Labov’s work with Harlem street gangs identifies such individuals as ‘Lames’. Being an outsider also has linguistic consequences which are usually identified as a lower frequency of non-standard fofms. Aspirers are individuals with ambitions beyond their immediate social domain. Studies have shown there to be a positive correlation between social ambition and non-local 27

forms (see Douglas-Cowie’s 1978 work in Articlave, and Mees & Collins’ 1999 work in Cardiff).

Examination of the role of the individual as innovator has focussed both on the type of person the innovator is and the type of relationships the innovator has. Using a methodological distinction between speaker and system, Milroy (1992: 169) differentiates between innovation and change, and the role of the individual as innovator is clear:

[ijnnovation and change are not conceptually the same thing: an innovation is an act of the speaker, whereas a change is observed within the language system. It is speakers, and not languages, that innovate.

Attempts to create a social characterisation of the individuals responsible for the actuation of change have led to what Milroy (1992: 172) describes as the ‘idealized speaker-innovator’. Labov describes speaker-innovators who lead sound change as those with the highest social status in the local community as measured by a social class index, or the speakers with the largest number of social contacts within the neighbourhood, yet who at the same time have the highest proportion of their acquaintances outside the neighbourhood. Such individuals, who have many ties within the community but who simultaneously have many outside links, have prestige both inside and outside the community. However, Milroy (1992: 173) argues that these two kinds of prestige are often in conflict. The individual is characterised more through the social ties that exist between individuals in Milroy’s view. As, according to Milroy, it is weak ties and not strong ties that are crucial in the diffusion of innovations, Labov’s speaker-innovator corresponds more to what Rogers and Shoemaker define as an early-adopter (in Milroy 1992: 184). Such individuals are central to the group, but their role is to diffuse the innovation to the group as a whole once it reaches them via the innovator, whose weak links to more than one group allow the individual to act as a bridge. Identification of the individuals responsible for the introduction of innovations into a speech community, as opposed to the individuals who adopt the innovation once introduced, may never be possible. However, attempts to procure profiles of the individual speakers who make up 28 the group so defined should be made, as a focus solely on the group can ‘make it difficult to hear the particular voices, and harder to take them seriously as part of the explanation for variation’ (Johnstone & Bean 1997: 235). Profiles of the speakers of the sample found to use the highest proportion of innovations in the data will be considered in Chapter 6.

2.2 Linguistic change

One of the self-evident facts about language is that it is continuously changing, and one of the most debatable and debated questions in linguistics concerns the causes and effects of language change. The axiom underlying much of sociolinguistic research is that at any given time a language, or a dialect, is variable. Variation can be observed geographically and socially, as noted in section 2.1. Language can also vary depending on context and situation. Prior to Labov’s early work, linguistic change was thought to be too slow to be observable synchronically (Bloomfield 1936). Labov’s work, and variationist work since, however, have shown that the orderly heterogeneity of language is of crucial importance in allowing us to detect change in progress in the study of variation.

Linguistic change can be interpreted as either system-based or speaker-based. System-based changes are endogenous changes brought about by pressures internal to the linguistic system. An example of linguistic change interpreted with system-based factors can be seen in the Northern Cities Chain Shift found in northern cities of the United States such as Chicago, Detroit, Cleveland and Buffalo. This chain shift, which has been investigated by a number of researchers, most notably Labov (1994), is a co-ordinated rotation of the mid and low vowels. Many recent findings on linguistic change in BrE, however, appear to be best accounted for by factors external to the linguistic system. The present work also considers the linguistic changes in progress suggested by variation in the data in the light of speaker-based factors which are external to the linguistic system. It is not within the scope of this work to detail all theoretical models of the causes and effects of language change (see further Weinreich, Labov and Herzog 1968, Milroy 1992, McMahon 1994, Labov 1994, 2001 and forthcoming, 29 for example). However, in this section a brief outline is offered of some of the processes and mechanisms of linguistic change which are seen as external to the linguistic system and are relevant to the findings of the present study.

2.2.1 Change within a dialect contact framework

The importance of mobility and face-to-face contact between speakers of different varieties in language change has been highlighted most clearly in Trudgill’s Dialects in Contact (1986). Unlike a situation of language contact in which change is an aid to mutual intelligibility, in a situation where speakers of two or more mutually intelligible dialects come into contact, why this should prompt change of any sort is not immediately obvious. One theory which goes some way to explaining the linguistic modification which can occur in a dialect contact situation is the accommodation theory (Giles and Powesland 1975, Giles and Smith 1979). Such a theory proposes that accent convergence, a reduction in pronunciation dissimilarities, will occur if a speaker wishes to gain the addressee’s approval. Conversely, the reverse process, accent divergence, is a strategy speakers can use to dissociate themselves from or show disapproval of their addressee. In dialect contact situations, linguistic accommodation is the norm. The accommodation is driven by social psychological factors rather than necessity, however. As speakers reduce dissimilarities when accommodating to other speakers, rather than imitate their addressee, an amount of imperfection in accommodation is likely (Trudgill 1986). Precisely which forms are likely to be modified in accommodation will be explored in section 2.3.

Within this dialect contact framework, two processes of language change are identified, dialect (or accent) levelling and diffusion Unprecedentedly rapid changes have been witnessed in spoken BrE in recent years, most of which appear to be best accounted for in the literature by a dialect levelling model of change (Cheshire, Gillet, Kerswill & Williams 1999, Williams & Kerswill 1999, Watt & Milroy 1999, Watt in press, Dyer 2000, in press). Dialect or accent levelling involves ‘the eradication of socially or locally marked variants which follows social or geographical mobility and resultant dialect contact’ (Watt & Milroy 1999: 26). This results in the reduction of the number of variants of 30 particular phonological, morphological or lexical variables available to speakers (Kerswill & Williams 1997: 214). Dialect levelling can be seen as ‘a linguistic reflex of the large-scale disruption, endemic in the modern world, of those close- knit, localised networks which can be shown to maintain highly systematic and complex sets of socially constructed linguistic norms’ (Watt & Milroy 1999: 26). Thus, in a mobile population, individual acts of accommodation replicated throughout the community can lead to the disappearance of marked local variants and the increased use of variants with a wider geographical and social usage and can thus lead to permanent language change. The process of levelling has recently been investigated in detail in varieties of BrE, notably by Watt in Newcastle (1998, in press) and Cheshire, Gillett, Kerswill & Williams in Milton Keynes, Reading and Hull (1999).

The other related process is that of geographical diffusion, a process whereby linguistic forms and innovations are diffused from one geographical area to another. This geographical or spatial diffusion often constitutes a kind of reflection of other types of diffusion. Diffusion can be described as sociolinguistic, in that linguistic changes are spread from social group to social group; lexical, in that changes are spread from word to word; linguistic, in that changes are spread from one linguistic environment to another; or spatial, in that changes are spread from place to place (Chambers & Trudgill 1980). For our purposes, the geographical diffusion of forms is what is of interest. The diffusion of forms across geographical space is sometimes differentiated into relocation diffusion and expansion diffusion. Relocation diffusion involves innovations being carried by speakers migrating to new locations. Expansion diffusion involves innovations being passed on through day-to-day contact between speakers who have adopted the innovation and those who have not (Gerritsen 1987). The processes of contact and accommodation are argued to be the same in each case, however (Britain 2000: 61). A gravity model of diffusion predicts and ranks the influence of a place on another place or places. The likely influence two places would have on each other is predicted by assessing and calculating distance and population factors. Social, psychological and physical factors must also be taken into account, however, as the speakers’ perceptions of distance can be minimised or maximised by such factors, and this can have an effect on the 31

influence of one place on another, as noted in section 2.1.1 (see further Britain 2000). Rather than a wave model with forms radiating out from a focal area reaching locations which are physically closer than those at greater distances, forms are thought to descend an urban hierarchy of ‘large city to city to large town, to town, village and country (Britain 2000: 62). Transportation networks and employment patterns are such that interaction between urban centres is likely to be greater than that between urban and nearby rural locations. Hierarchical effects have been found by a number of researchers (see for example Trudgill 1983, Hernandez Campoy 2000).

Although, in the current British literature, a distinction is not always drawn between levelling and diffusion, as the term levelling can be given a broader definition which includes the reduction of variants and the emergence and adoption of variants in a variety (Williams & Kerswill 1999, Foulkes & Docherty in press), the present study chooses to differentiate the two processes. Despite the slight differences in terminology in the literature, it is agreed that the result of both processes is homogenisation. Large-scale homogenisation of BrE has been the focus of much research in recent years.

In most cases the homogenisation is seen as being contact-induced, as already noted. Crucially, this contact is necessarily face-to-face interaction between speakers. Precisely which speakers introduce new forms into a variety can perhaps never be known. Mobile speakers who have weak ties to the community are likely to be instrumental in the introduction of new forms. It has been suggested that new forms can be introduced through language missionaries (Steinsholt, cited in Trudgill 1986: 57). Such individuals move away from their native area for a time and then return, presumably having acquired new forms. As the speakers are insiders, their language is not rejected as that of outsiders might be (Trudgill 1986: 56). Another possible source of new forms is through the media. It has been assumed in the past that media influence can only reinforce the effects of the accommodation which occurs in face-to-face contact (Trudgill 1986: 40-1). Trudgill argues that if the mass media were the source of diffusion and linguistic change, the whole of Britain would be influenced by an innovation simultaneously, which is not the case. Recently, interest in the role of the mass 32 media in transmitting change has increased (Stuart-Smith 2001), particularly given the increase in the appearance of non-standard, mainly southern, accents in youth-oriented mass media broadcasts. However, if we were to find evidence which suggested that the media were responsible for the transmission of innovations, we would still not know how the innovation had moved from the television to the mouths of members of the speech community, that is, which speakers initiate the changes in being the first to adopt an innovation. The question of the precise processes involved in the actuation and embedding of innovations leading to language change posed by Weinrich, Labov & Herzog in 1968 remain largely unanswered.

An innovation can only be called a change when it becomes embedded in the linguistic system after having been transmitted to some critical number of individuals who have evaluated the new form favourably. It would seem that positive evaluation of forms is crucial to their adoption. This may well entail the attachment of some socially meaningful association to the form. Attempts to access and investigate such overt or covert evaluations is the topic of the next section.

2.3 Ideology and evaluation

2.3.1 Language ideology

Although various terms can be used, with slightly differing emphases signalled, to refer to the concept of ideology and its application to linguistic investigation (linguistic ideology, language ideology, ideology of language), for the purposes of the present work the term language ideology will be used. The concept has not been delimited definitively, and it is not within the scope of the present work to detail its use in sociolinguistics, anthropology and cultural studies (see Woolard 1992 for further discussion). However, language ideologies have been defined as ‘sets of beliefs about language articulated by users as a rationalization or justification of perceived language structure and use’ (Silverstein 1979 in Woolard and Schieffelin 1994). 33

The basic division in studies of ideology, according to Woolard & Schieffelin (1994), is between the neutral and critical values of the term. The neutral value views ideology as a neutral semiotic process (Silverstein 1992, 1995). The critical value views ideology as a strategy for maintaining power and is often used in studies of language politics and language subordination (Lippi-Green 1997). The neutral value is of interest to the present research.

Irvine and Gal (2000: 35) argue that the ideological aspects of linguistic differentiation emerge as a consequence of attempts by individuals to formulate understandings of linguistic variation which can be mapped onto significant people, events and activities.

As part of everyday behavior, the use of a linguistic form can become a pointer to (index of) the social identities and the typical activities of speakers. But speakers (and hearers) often notice, rationalize, and justify such linguistic indices, thereby creating linguistic ideologies that purport to explain the source and meaning of linguistic differences....Participants’ ideologies about language locate linguistic phenomena as part of, and evidence for, what they believe to be systematic behavioural, aesthetic, affective, and moral contrasts among the social groups indexed (Irvine and Gal 2000: 37).

Importantly, as Milroy (2000: 9) notes, people, events and activities viewed as significant will vary between communities. Furthermore, they may vary within communities as changes in reactions to saliences of locations may alter and attitudes towards salient social groups may shift. This is an important point to which we will return in Chapter 6.

Irvine and Gal (2000: 36-38) identify three semiotic processes by which ideologies recognise (or misrecognise) linguistic differences, how they locate, interpret, and rationalise sociolinguistic complexity by identifying linguistic varieties with ‘typical’ persons and activities, accounting for the differentiations among them. These three processes are termed iconization, which involves a transformation of the sign relationship between linguistic features and the social images with which they are linked, fractal recursivity, which involves the projection of an opposition, salient at some level of relationship, onto some other 34 level, and erasure, which is the process in which ideology, in simplifying the sociolinguistic field, renders some persons or activities (or sociolinguistic phenomenon) invisible.

Silverstein (1992, 1995) views ideology as a system for making sense of the indexicality inherent in language: in other words, how language forms index speakers’ social identities. This indexicality, according to Silverstein, can be ranked into different orders of generality. First order indexicality involves an association of a linguistic form with some socially meaningful category. Second order indexicality is a metapragmatic concept which involves the overt or covert noticing, discussion and rationalisation of basic first order indexicality. Hence ideology can be identified through metalinguistic discourse. Language ideologies thus entail the selective association of a linguistic form with some meaningful social group. According to Milroy (2001), they become visible in style shifting in careful speech, hypercorrections, hyperdialectisms, and beliefs about language which are resistant to rational reflection.

Woolard (1992: 240) argues ‘in even the most correlational sort of sociolinguistics, from the early important work of Labov, the motive force of linguistic change lay in conceptions and evaluations of language. These surely fit under the broad banner of “ideological”’. In correlational studies a direct relation between linguistic variation and some aspect of social differentiation is often suggested. However, ‘the correlation is in fact mediated by an ideological interpretation of the meaning of language use’ (Woolard 1992: 242). Indeed, according to Irvine and Gal (2000: 47) a situation of language change as an ideologically fuelled process of increasing divergence is found in one of the earliest sociolinguistic studies. Labov’s (1963) Martha’s Vineyard study demonstrated how contrasts among ethnic groups of islanders (Portuguese, Indians and Yankees) in the 1930s were replaced by a contrast between islanders and mainlanders in the 1960s. The development of the tourist industry made the latter contrast more socially significant than the local, intra-island contrasts. Thus

2 Labov himself (1979: 329) discounts the power of ideology to affect speech forms, holding the view of ideology as overt political discourse (Woolard 1992: 240). 35 the islander phonology diverged more sharply from the mainland forms. Although Labov did not explicitly invoke a language ideology framework to explain the motivation for the language changes in evidence, its relevance appears central to an interpretation of the motivation of the sound change observed.

Recently, a framework of current work on language ideology has been used by Milroy (2000) to compare the ways in which language varieties are evaluated in Britain and the United States as standard and non-standard. Milroy’s claim is that language varieties in Britain and the United States are differently ideologised so as to foreground social class groups in Britain and racial and ethnic groups in the United States (2000: 24). In Britain, phonology is critical, with RP, a class- marked accent, treated as a central reference point, but imagined as elite rather than mainstream. The class distinction is further marked with the stigmatisation of urban varieties spoken in industrial cities and associated with WC speakers. In the United States, a mainland levelled variety shorn of ideologised grammatical and lexical forms which index social, regional or ethnic groups is the central reference point. Varieties associated with ethnic groups, most particularly African-Americans, are stigmatised. Thus, salient differences in language ideology which are ‘deeply embedded culturally, cognitively and historically’ (Milroy 2000: 26) are in evidence in two cultures which share a common language.

A language ideology framework has recently been used expressly to interpret findings of variationist studies and to explain the motivation for linguistic change (see Anderson and Milroy 1999, Anderson 2000, Milroy 2001, Dyer 2000, in press). In a study of vowel shifts in African-American English (AAE), Anderson (2000) demonstrates how local ideology structures the direction of recent phonological change in Detroit AAE. In the United States, system internal factors are customarily invoked to explain such changes as the Northern Cities Shift (Labov 1994). However, certain ethnic groups, such as African-Americans, appear not to participate in such sound changes. Anderson suggests that social factors are implicated in the Southern Shift, in that participation in the Southern Shift of African-Americans resident in the South has been inhibited ‘by a series 36 of ideologized linguistic distinctions between themselves and southern White speakers’ (Anderson 2000: 4). However, when the African-Americans migrate north the distinction between African-American and southern White speakers is no longer relevant. The relevant ideological opposition for the African- Americans in Detroit is with Anglo Northern speakers. Thus, the vowel system is liable to drift in a manner consistent with the principles outlined by Labov (Anderson 2000: 4). Phonological changes, such as loss of conditioning environment for /ai/ monophthongization, indexes the opposition between African-Americans who migrate to Detroit from the South and Anglo Northern speakers. This simultaneously allows Detroit African-Americans to align with southern White migrants in Detroit. A system emerges in which the North/South dichotomy is indexed, but the southern Black/southern White dichotomy is erased and appears no longer relevant. Thus, Anderson (2000) and Milroy (2001) argue that the direction of an internally motivated or more general type of change such as levelling can be structured by local ideology.

Similarly, ethnic affiliations and shifting ideologies are demonstrated in Dyer’s study of Corby (2000). Corby, a former steel town in the English Midlands, was the site for large numbers of steelworking in-migrants from Glasgow in the 1960s. Dyer’s study investigated whether a new dialect had formed in Corby as a result of the in-migration from Scotland and how far the new dialect had been influenced by Scottish English spoken by the newcomers. A quantitative analysis of the reflexes of phonological variables characteristic of Scottish English across three generations was undertaken to reveal whether characteristically Scottish variants had been adopted by the younger speakers of the sample. This was indeed found to be true, in some cases, for example, the Scottish monophthongal form [o:] for the GOAT vowel was used by the young males of the sample. By utilising a language ideology framework and investigating attitudinal information. Dyer argues that some historically Scottish features have been reallocated in the Corby dialect (2000: 164). Use of such variants indexes a local rather than a Scottish identity to the young speakers. A realignment of social groups has occurred in Corby, and use of historically Scottish forms by young speakers indexes a Corby identity in opposition to the neighbouring town of 37

Kettering. Thus a shift and reallocation of perceptions of salient social categories is in evidence from ethnic group to town community. Such insight would not have been accessible, Dyer argues (2000: 168), through use of a traditional quantitative study which does not analyse and correlate the metapragmatic remarks about language and ethnicity from a language ideology perspective. Such a language ideology framework will be utilised in the present study with informants’ opinions and attitudes examined as instantiations of second order indexicality. The saliences of social groups will be considered in these instantiations of second order indexicality.

2.3.2 Evaluation and salience

Finally, we consider aspects of evaluation and linguistic salience which are relevant to the present work. Overt evaluation of forms by informants and covert evaluation as evidenced in style shifting in more monitored speech are clearly instantiations of second order indexicality and thus manifestations of speakers’ language ideologies. However, brief consideration of the concepts of prestige and salience as used in work on linguistic variation and change is necessary before we turn to the particulars of the present study.

If we begin with the notion of prestige, Trudgill (1974: 96) first proposed the notion of covert prestige to define the opposing set of values that allow localised or non-standard features stigmatised by dominant social classes to be maintained and even to expand in the face of such stigmatisation. This covert prestige is in opposition with overt prestige, which is the ‘prestige of the high-status group representing, symbolically, the whole community’ (Hudson 1980: 201). As noted in section 2.1.2, female speech has frequently been associated with high-prestige forms. The vernacular/standard continuum, however, has come to be less relevant than a local/non-local opposition, with female speech increasingly associated with supra-local forms. Moreover, Milroy, Milroy and Hartley (1994) argue that female use of a variant, even a traditionally stigmatised variant such as the glottal stop, may bring about a change in the prestige attached to that variant, 38

female patterns of use are hypothesised to be instrumental in bringing about a reversal of the traditional low evaluation of the glottal stop. The generalisation suggested here is not that females favour prestige forms, but that they create them; i.e., if females favour certain forms, they become prestige forms (Milroy, Milroy and Hartley 1994: 26).

In more recent work which investigates patterns of variation and change in the vowels of Newcastle English within a dialect levelling framework, Watt and Milroy (1999: 42-43) find the notions of prestige and stigma inadequate to explain the social motivation for the increase in the mainstream northern monophthongal forms in the variety. Rather, ‘[t]he correct generalisation appears to be that men tend to retain localised forms (which may or may not be stigmatised) and women adopt supra-local forms (which may or may not be prestigious) (Watt and Milroy 1999: 43).

Style-shifting appears to be clearly linked to notions of prestige. According to Labov (2001: 196), ‘social awareness of a given variable corresponds to the slope of style shifting’. Labov differentiates between change from above and change from below. Changes from above, according to Labov (2001: 274) take place at a relatively high level of social consciousness, show a higher rate of occurrence in formal styles, and are often subject to hypercorrection. Change from below, on the other hand, involves changes below the level of social awareness. Furthermore, Labov (2001: 196) differentiates between markers, indicators and stereotypes. Changes from below begin as indicators of variation, forms which are stratified by age, gender and social class, but forms for which no social awareness is shown. As they proceed, they acquire social recognition and become markers of variation. This recognition is generally in the form of social stigma and the forms become subject to style shifting. Ultimately they become stereotypes and are the subject of overt comment ‘with a descriptive tag that may be distinct enough from actual production that speakers do not realize that they use the form themselves’ (Labov 2001: 196-7).

However, such differentiations between forms and types of change do not explain why speakers should be more aware of some variables than others. Despite a degree of circularity in the argument, it is suggested that greater 39 awareness attaches to forms which are overtly stigmatised in a community. According to Trudgill (1986: 11), the salience which attaches to markers is what turns variables into markers in the first place. Salience is due to stigmatisation which is often through the existence of a high status variant of the stigmatised form which tallies with the orthography, linguistic change, phonetic distance and phonological contrast. Trudgill (1986: 37) claims that during accommodation it is the salient features of the target variety that are adjusted to. A number of factors can combine to delay modification, however. These are phonotactic constraints, homonymic clash and extra-strong salience (which attaches to features which are overly strong markers of the dialect being accommodated to).

Kerswill, who defines the notion of salience as pertaining to ‘a linguistic item or feature that is in some way perceptually and cognitively prominent’ (2000: 1), argues that there is no way of predicting which salient features will be adopted and which avoided except by having recourse to non-linguistic information. Features should therefore be viewed in terms of their social patterning as well as their salience, as salience, however defined or caused, will be different for different social groups. Language-external cognitive, social psychological, pragmatic or interactional factors are those which motivate speakers to behave in a certain way (Kerswill 2000: 19).

Features which can be said to be salient also contribute to the non-linguist’s identification of accents and dialects. A recent investigation of dialect perception and its relation to speech community focussing was undertaken by Kerswill and Williams (2000). As part of their project on The role o f adolescents in dialect levelling, Kerswill and Williams presented their informants with taped samples of ten speakers chosen to be locally relevant. What is of interest to the present work is that the samples presented to the informants in Hull contained a speaker from Middlesbrough. The Middlesbrough accent was incorrectly identified by all the judges. According to Kerswill and Williams this suggests a iack of familiarity with the accent, due, probably, to a minimal amount of contact between that town and Hull’. Interestingly, the largest number of erroneous identifications of the accent by both the MC and the WC judges involved the citing of Liverpool as the origin of the speaker, which we shall consider further 40 in section 3.1.1.2. Misidentification of the accent as Newcastle achieved the second highest score. Kerswill and Williams cite as possible features leading to the misidentification intonational features, use of glottal reinforced pre-vocalic N and realisations of the GOAT and f a c e vowels (half-close monophthongs which are similar to those of Newcastle English) and the n u r s e vowels (an [§:] realisation which is similar to that of Liverpool English). Features which are above and below the level of awareness and how those above are evaluated by informants are also relevant to the present research and we shall consider them further in Chapters 5 and 6.

With the theoretical and contextual background in which the present study is grounded in mind, we turn to the topic of language variation and innovation in Teesside English. We begin with details of the design of the study, which is the subject of the following chapter. Chapter Three The Design of the Teesside Study

3.0 Introduction

Any empirical research undertaken within the social sciences must make explicit how data have been collected, analysed, presented and interpreted. Such a principle of accountability is central to the discipline of sociolinguistics. Justification for decisions made in the design of the study and the explicit elucidation of methods utilised in the research allows other researchers to assess the analysis of data and the interpretation of results. This chapter presents an account of the design of the Teesside study before analyses of the results and interpretations of the findings are offered in following chapters.

The Teesside study is based on an experimental model. Such a model describes ‘any method which entails the direct manipulation of an informant’s responses’ (Milroy 1987b: 5). An experimental study is one in which the investigator has control over the data accessed, one in which ‘a native speaker is used to provide specified types of fact about the target language or variety’ (Milroy 1987b: 5). This is in contrast to an introspective model, where the description is based on self-observation, and an analytical model where generalisations are based on a corpus of independently collected data. Decisions made in the design of such an experimental sociolinguistic study determine, to some extent, the outcome of the research and are thus of crucial importance. The decisions are determined by experimental hypotheses made before the data are collected. Hence the present study has been designed in light of the research questions it addresses as outlined in Chapter 1.

Hudson (1980: 144) lists five methodological stages in a sociolinguistic study, stating that ‘[m]ethodology is both important and problematic at all stages’. These stages are: 42

A. selecting speakers, circumstances and linguistic variables; B. collecting the texts; C. identifying the linguistic variables and their variants in the texts; D. processing the figures; E. interpreting the results.

This chapter is concerned with the first two stages listed. Stages C and D are examined in the following chapter. The final stage, the interpretation of the results, will be examined in Chapters 6 and 7.

We begin, in section 3.1, with an overview of the sampling decisions made in the study, including justification for the choice of locality for the research, the choice of independent social variables used, and the design of the fieldwork sample. We next turn to the method of data elicitation utilised in the present study in section 3.2, and consider the unique methodological setting underlying the design of the innovative method of data elicitation which the Teesside study is trialling. Finally, particulars of the fieldwork are considered in 3.3, which includes information on access to and biographical data of the informants whose sample of speech was analysed in the study.

3.1 Sampling Decisions

According to Sankoff (1980) in her discussion of quantitative methodology, in order to obtain good data the researcher has three different kinds of decision to make about the sampling procedures to be employed in the study. These are:

1. To define the sampling universe. That is, to delineate, at least roughly, the boundaries of the group or community in which one is interested. An adequate sample frame which investigates group members may then be sought. 2. To assess the relevant dimensions of variation within the community - this involves constructing stratification for the sample. Thus, we must ask whether ethnic group, sex, or social class of speaker might affect the kind of language used. Most studies so far have shown that to a very great extent they do, as does situational context. 3. The sample size needs to be fixed. 43

After decisions 1, 2, and 3 have been made, Sankoff notes that sampling may proceed through use of random methods (cf. Labov 1972), or through a social network approach (cf. Milroy 1987a). This section will consider first how the sample universe is defined, with reasons given for the choice of locality of the study. The relevant dimensions of variation within the community are then considered, and the selection of independent social variables is outlined. The sample size and the design of the fieldwork sample will then be presented.

3.1.1 Location

The sampling universe of the present study is Teesside in the North East of England. The town of Middlesbrough, being the most densely populated urban centre on Teesside, can be seen as delineating the boundaries of the community under investigation (see Figure 3.1 for geographical location). Within Teesside, Middlesbrough has a population density per hectare of 26.7 (population 144,100), compared with 9.7 in Flartlepool (population 90,700), 5.6 in Redcar and Cleveland (population 138,100), and 8.7 in Stockton-on-Tees (population 178,300) (Tees Valley TEC 2001).

Figure 3.1: Geographical location of Middlesbrough 44

The choice of Middlesbrough as the location for the study is motivated by six factors. The first is that Middlesbrough English (MbE) represents an urban variety of British English (BrE) which has not hitherto been investigated, as noted in Chapter 1. The second relevant factor is the history of the town, which has consequences for the development both of MbE and of the identity of Middlesbrough as an urban centre. The third related factor is the geographical position of Middlesbrough. This has relevance both in terms of the identity of the urban centre, and in terms of recent variationist studies of nearby urban centres with whose findings the present study is compared. Historical and geographical factors, and the image of the town in popular culture, are relevant to the shifting identity of Middlesbrough. This shifting identity is the fourth, and perhaps the most important factor in the selection of Middlesbrough as the locality of the study. The fifth factor is the position of the town in the current economic and social climate. And the final consideration is the fact that I am a native of Middlesbrough. The relevance of each factor to the present study will now be considered in turn.

3.1.1.1 Lack of previous studies

Existing literature on the variety in question is sparse, so much so that, to my knowledge, only one study of the Cleveland dialect exists. This dates back to 1868 and is largely a glossary of the dialect (Atkinson 1868). For comparative purposes, it is of limited use given the synchronic investigation of social variation which is the focus of the present study. The other available source of data for real time comparative analyses is the Survey of English Dialects (SED) (Orton & Dieth 1962-1971). The localities featured in the SED which lie closest to Middlesbrough are Stokesley (area Y2), which lies approximately 15 miles south, Skelton (area Y3), which lies approximately 18 miles east, and Bishop Middleham (area D5), situated approximately 20 miles north. However, the SED was diachronic in orientation, as noted in section 2.1.1. Its design was planned in the belief that ‘it is amongst the rural population that the traditional types of vernacular English are best preserved today’, and further that ‘in this country men speak vernacular more frequently, more consistently and more genuinely than women’ (Orton and Dieth 1962: 14). The sampling decisions of the survey 45 reflected these beliefs: the majority of informants used in the SED were male and over 60 years of age, and the majority of localities were rural. All informants from the three rural localities used in the SED closest to Middlesbrough were over 50 years of age and were recorded between 1951 and 1953. Given that the current study is a synchronic investigation into an urban variety of BrE, particularly a relatively recently developed urban variety of BrE (which will be considered in the following section), for comparative purposes these data are also of limited usefulness. With little existing literature on the variety in question, then, an investigation into the English spoken on Teesside is long overdue.

Additionally, a synchronic study of Teesside English is particularly useful as a contribution to the current state of knowledge of BrE varieties. Middlesbrough is well placed geographically as the locality of a study which seeks to gain insight into the extent of the spread of certain current vernacular changes in BrE. A number of recent studies of urban centres in the North and North East of England have been undertaken, the findings of which will be useful for comparative purposes (see further in section 3.1.1.3). A study of Teesside English is also timely in view of the interest in the development of supra-local or regional standards and particularly of a North Eastern regional standard (Watt and Milroy 1999, Watt in press).

3.1.1.2 History of Middlesbrough

The history of Middlesbrough as an urban centre can be said to begin in 1830 with the arrival of the railway. Prior to that date, agriculture dominated the economy and society of Middlesbrough, which consisted of only four houses and 25 inhabitants in 1801, and only 40 inhabitants in 1821 (Moorsom 1993). In 1825 the first steam-powered public railway in the world was opened between the nearby towns of Stockton and Darlington. Its purpose was the conveyance of coal from the pits in County Durham to the South. However, to avoid the navigational hazards of the River Tees, it was decided to extend the railway from Stockton to Middlesbrough, which lies six miles nearer the sea Where the water is deeper. Middlesbrough, at the time, consisted of ‘500 acres of bleak salt marshes’ (Briggs 1996: 2). In 1829, Joseph Pease of Darlington and five business 46 associates, who became known as ‘The Owners of the Middlesbrough Estate’, purchased the Middlesbrough Farm for the sum of £30,000. Pease’s vision of the Middlesbrough of the future, as indicated by his 1828 diary entry: ‘imagination here has very ample scope in fancying a coming day when the bare fields ... will be covered with a busy multitude and numerous vessels crowding to these banks denote the busy seaport’, was soon to be realised, as ‘[n]o other town in Britain could match the spectacular demographic and economic growth that transformed a small collection of farms into one of the nation’s leading industrial centres’ (Taylor 1996: 53).

The phenomenal growth of Middlesbrough, the world’s first railway town, is seen as having about it ‘a mythical quality, both heroic and also tragic’ (Taylor 1996: 53). The Owners of the Middlesbrough Estate were initially responsible for the planning of the proposed new town and some 30 acres sub-divided into 125 lots were set aside for its development. A broader industrial base than the initial coal-exporting industry was necessary for the speculative business venture of the development of Middlesbrough to succeed. In order to ensure a sound return on their investment the Owners of the Middlesbrough Estate encouraged the establishment of new industrial ventures in the town, and in 1841 Bolckow and Vaughan’s Middlesbrough Iron Works came into operation. The discovery of ironstone in the Cleveland hills in 1850 ensured that within a mere 40 years Middlesbrough had become the biggest producer of pig-iron in the world (Chase 1995). This ‘boom time’, with iron prices, company profits, and population levels all at unprecedented highs, led ‘the more recent recruits to the iron-making capital of the world to view Middlesbrough’s market dominance as natural and impregnable’ (Nicholson 1996: 33). The tremendous growth in industry led W.E. Gladstone, on his visit to Middlesbrough in October 1862, famously to declare ‘[t]his remarkable place, the youngest child of England’s enterprise ... is an infant, gentlemen, but it is an infant Hercules’ (in Briggs 1996: 1).

Such tremendous growth in industry entailed an equally tremendous increase in the workforce, and the population growth in Middlesbrough in the nineteenth century was quite remarkable. Although other cities in the United States and Australia grew from nothing even faster during this time, no other British town 47 exceeded the expansion of Middlesbrough. Seen as the largest and the most impressive of the new towns of the nineteenth century (Chase 1995:3), the steady growth in population meant that, by the turn of the century, Middlesbrough was a boom town with a population of over 91,000 (see Figure 3.2).

Figure 3.2: Census figures 1801-1901

100000 91,302 90000 80000 75,532 70000 60000 55,934 50000 39,284 I population 40000 30000 18,892 20000 5,463 7,631 10000 25 35 40 154 0 —b

o CN CO LD CD OO CD o 00 00 00 co oo OO OO OO OO CD

Throughout the nineteenth century, people flocked to Middlesbrough looking for work, not only from the town’s rural hinterland but also from further afield, especially from Durham, Staffordshire, South Wales, Scotland and Ireland. In 1861, when the population was recorded as 18,892, 73.2% of the population had been born in the county in which Middlesbrough was situated, Yorkshire. By 1871, however, when the population was recorded as 39,284, the native county population had sunk to 50.1% (Briggs 1996: 8). Immigrants from the United States and the East Indies were also attracted to Middlesbrough, with 551 people from ‘foreign parts’ recorded in the 1871 census. Limited employment opportunities existed for women in the male dominated town, however, and of all the urban centres with a population of at least 50,000 in England and Wales, Middlesbrough had the lowest level of female employment at that time (Taylor 1996: 61).

Middlesbrough thus had a very different demographic pattern from that of most other British towns. Frequently compared with frontier towns, it was stated that Middlesbrough’s ‘rapid growth, the heterogeneous composition of its population, 48 and the preponderance of the male sex, recall features generally credited only to towns of the American West’ (Ravenstein cited in Briggs 1996: 8). There are comparisons to be made between Middlesbrough and the new American townships. Indeed, Chicago came into existence as a planned settlement in exactly the same year as Middlesbrough. Conditions were quite different, however, as ‘ [t]he establishment of urban centres in frontier America was based upon an abundance of land rather than, as in Middlesbrough’s case, commercial and industrial enterprise on small plots of bought land’ (Turner 1996: 98). An average population of two people per square mile is recorded on the early American frontier. A comparison of this figure with even the very beginnings of the growth of Middlesbrough (in 1831 there were almost 200 persons per square mile, which had risen to 3,540 by 1841) suffice to illustrate the difference (Turner 1996: 98).

Life was hard in the cramped and unsanitary conditions which the pioneers of Middlesbrough suffered. With a death rate well above the national average in the second half of the nineteenth century, Middlesbrough was ‘one of the most unhealthy towns in the country’ (Taylor 1996: 60). Overcrowding and poor housing were the conditions encountered by the constant stream of migrants. It was reported by the Poor Law Relieving Officer for the Middlesbrough district in 1867 that ‘hundreds of children (of ironworkers) are running about bare legged without any education; they are mostly Irish and Welsh’. Such conditions were perhaps inevitable in an environment of rapid growth which was not without periods of depression, particularly that between the mid-1870s and the 1880s which pre-empted the transition from iron to steel production. Indeed, ‘Middlesbrough’s utilitarian appearance resulted from its economic achievements as the country’s leading iron and steel producing centre - the noise, the dirt, the poverty, the ill-health, and the sacrifices to be made were, it was argued, the price to be paid for progress’ (Turner 1996: 82).

From the 1840s there was growing Irish migration to Middlesbrough. Large- scale Welsh migration to Middlesbrough took place in the period 1851-1871. Most Irish immigrants arrived as single males and were largely unskilled (Chase 1995). By contrast, the Welsh of Teesside tended to be skilled and moved to the 49

area in family groups (Lewis and Ward 1995). By the 1870s, the Irish were outnumbering the Welsh by three to one and, as one in five adult males in Middlesbrough was Irish, Middlesbrough was second only to Liverpool in terms of the size of its Irish population in the nineteenth century. However, the Welsh were assimilated into the wider community of Middlesbrough more smoothly than the Irish. Such was the assimilation that by 1910 members of the exile Welsh community represented both of the Teesside borough constituencies. This was seen as evidence ‘less of the success of that community in sustaining its identity than of the extent to which it had been absorbed by the locality’ (Lewis and Ward 1995: 569). Part of the difference in assimilation was based on religious affiliation, which was a key element in the evolution of the political culture of Middlesbrough. The Welsh were predominantly Nonconformist. The Irish, on the other hand, were predominantly Roman Catholic. Such was the extent of the Catholic population of Middlesbrough by the 1870s that the Catholic Church made Middlesbrough the centre of one of its new dioceses for the North of England (Lewis 1996). There were incidents of anti-Irish activity from time to time. Nonetheless, the Irish may have been assimilated into the Middlesbrough community with greater ease than they were in other urban centres with high Irish migration, as ‘sustained hostility does not seem to have developed on the same scale as in Liverpool or Glasgow’ (Lewis 1996: 112). In fact, although the Welsh can be seen to have adapted and advanced socially and economically in Middlesbrough, it can be argued that the Irish were ‘perfectly assimilated into the dominant popular culture of the region, a culture that was unskilled, non-unionised, hard-drinking and hard-living’ (Chase 1995: 6).

Such was the ‘melting pot community’ out of which MbE developed. As Lewis and Ward state with reference to the Welsh on Teesside, ‘[t]he language seems rarely to have survived into the second generation’ (1995: 570). A process of koineisation is likely to have occurred in the community. Koineisation, a dramatic form of dialect contact studied recently in Milton Keynes (Kerswill & Williams 1997) and the Fens of eastern England (Britain 1997), follows the mass settlement of a relatively sparsely populated area. According to Siegel’s definition, a koine is: 50

the stabilized result of mixing of linguistic subsystems such as regional or literary dialects. It usually serves as a lingua franca among speakers of the different contributing varieties and is characterized by a mixture of features of these varieties and most often by reduction or simplification in comparison (Siegel 1985: 363).

Although the legacy of the Welsh migration is still apparent in Middlesbrough today (an Eisteddfod is still held biennially in the town), the influence of the larger Irish migration into Middlesbrough can perhaps be felt in the similarities of MbE with Liverpool English. An example of this may be seen in the merger of the lexical sets NURSE and SQUARE (nurse pronounced [ne:s]). This is cited as a feature of Liverpool English (Wells 1982: 372 and Newbrook 1999: 95). It is also a feature which is seen as a shibboleth of MbE1.

In the first half of the twentieth century Middlesbrough continued to grow, although not at such a rapid rate as it had in the nineteenth century. Expansion of the industrial base continued, with chemical plants and oil refineries becoming operational along the Tees. The town’s post-War reconstruction and redevelopment were undertaken with the guidance of Max Locke’s commissioned ‘Middlesbrough Survey and Plan’. However, as the century progressed the heavy industries went into decline. In the second half of the century, shipbuilding on the Tees ceased and the docks were closed. In 1993 the Cleveland Iron Plant was closed bringing to an end the long tradition of ironmaking which had begun in 1854 (Moorsom 1995: 38).

As the industries went into decline, so too did the population of Middlesbrough. After increasing steadily since 1830, a plateau was reached in the 1960s. After 1971 the upward trend was reversed and a decline was observed in 1981 and 1991 (see Figure 3.3 below).

1 Many informants in the present study used examples which are illustrative of a NURSE / SQUARE merger in response to the second question in the Identity Questionnaire ‘Can you recognize the accent of Middlesbrough. If so, how?’ Additionally, some informants, when elaborating on this topic, mentioned the Irish influence on MbE. 51

Figure 3.3: Census figures 1911-1991

157,308157,313 160,000 147,336 150,430 138,960 140,850 140.000 I 131,170 120.000 ;104,767 100,000 I 80,000 | ] population 60.000 I 40.000 !

20.000 ' 0 .

CM CO LO CO 00 CD CD CD CD CJ> CD O) CD CT>

Attempts to diversify the local economy have been made, and some evidence of regeneration exists. However, a decline in population is still evident. The current economic situation will be considered in more detail in section 3.1.1.5. Aspects of the evolution of Middlesbrough’s political culture and its changing administrative bodies will be outlined in sections 3.1.1.3 and 3.1.1.4.

3.1.1.3 The geographical position of Middlesbrough

Middlesbrough’s geographical position has relevance to the present study on two levels. Its position on the North East coast of England is significant for comparative purposes with other recent studies, as noted in section 3.1.1.1, whilst its exact location along the River Tees is crucial to its identity, which is a central concern of the study.

Middlesbrough lies between two urban centres which have been the sites for major recent studies of language variation, Newcastle and Hull. Newcastle lies 38 miles (61 kms.) north of Middlesbrough, as noted in Chapter 1, and Hull is situated 88 miles (141 kms.) south (see Figure 3.4). 52

Figure 3.4: Location of Middlesbrough in the North of England

•►tyvlewcastle upon-Tyne • Durham Vv__ • Middlesbrough

Y o r k sh ir e

Kingston-upon-Hull ♦ Manchester

Newcastle is one of the locations of the study of language variation by Milroy, Milroy and Docherty (1997). It is also the location for a number of other recent studies, some of which are connected with the aforementioned 1997 research (Milroy, Milroy & Hartley 1994, Watt & Milroy 1999, Watt 1998, 2000a, in press, Beal 1999, Beal & Corrigan 1999, Trousdale 2000, Allen 2001). Hull is one of the three locations featured in the study of dialect levelling in BrE by Cheshire, Gillet, Kerswill & Williams (1999). Vowel fronting in Hull has also been investigated by Watt (2000b). Additionally, recent research has been undertaken in other nearby localities (see Watt & Tillotson 1999 for a recent study on Bradford English, and Watt & Ingham 2000 for a recent study on Berwick English). The data from Middlesbrough will be compared most closely with findings from Newcastle as reported in Docherty et al. (1997) and findings from Hull as reported in Cheshire et al. (1999).

Middlesbrough, as we have seen, owes its ‘birth’ in the nineteenth century and its very existence as an urban centre to its geographical position along the River Tees. Its position along the river has continued to have significance for its identity in the twentieth century. The River Tees has traditionally stood as a boundary between County Durham to the north and Yorkshire to the south, as noted in Chapter 1. Middlesbrough lies on the south bank of the River Tees. For much of its history, then, Middlesbrough has been situated in the North Riding of the County of Yorkshire. As noted earlier, several changes in local government administrative boundaries have been made in recent years, however. These changes and their implications for the identity of Middlesbrough and the 53 conurbation along the River Tees are considered in more detail in the following section.

3.1.1.4 The shifting identity of Middlesbrough

Middlesbrough has never been ‘a fixed entity’ (Pollard 1996: xi), and the shifting and fluid nature of its identity is central to the present study. This identity is viewed in terms of local government administrative boundaries; the identity of Middlesbrough as viewed by the informants of the study; the image of Middlesbrough in popular culture; and the placement of MbE in dialect groupings. In this section we consider the shifting local government identity of the town first. We then consider its position in terms of popular culture. Finally we consider where Middlesbrough lies in terms of dialect groupings. The informants’ views on the question of the identity of Middlesbrough are considered in Chapters 5 and 6.

Pre-industrial Middlesbrough was situated in the North Riding of the County of Yorkshire. Although Middlesbrough became a Municipal Borough in 1853 and a County Borough in 1889, it was still situated in the North Riding of Yorkshire until 1968 (see Figure 3.5).

Figure 3.5: Middlesbrough in the North Riding of Yorkshire

YORKSHIRE Middlesbrougl WESTMORLAND:

Sheffield NOTTS 54

The County Borough of Teesside was formed in 1968, after a Commission which had been ordered to review the organisation of local government recommended that:

[t]he present system under which eleven local authorities are responsible, in one way or another, for the administration of Teesside, has ceased to match the complex pattern of urban and industrial growth, the rapid increase and spread of population, and the main tasks confronting the authorities (in Moorsom 1996: 14).

Prior to that date, the River Tees had stood as a boundary between the County of Durham to the north and Yorkshire to the south, as noted. In 1968, with the formation of the County Borough of Teesside, the conurbation along the River Tees was drawn together, bringing the urban centres north of the Tees, Stockton- on-Tees and Billingham, which had previously been situated in County Durham, together with those on the south bank (see Figure 3.6 below).

Figure 3.6: Middlesbrough in Teesside

There were mixed local reactions to the change, despite the Minister of Housing and Local Government’s view that:

[t]he exciting industrial and urban growth in recent years on Teesside has drawn once separate townships closer together and it became increasingly clear that existing local authority boundaries were hindering further development for the community as a whole. 55

For local government administration purposes there was a need for unification of a number of local authorities. This is undoubtedly the most extensive amalgamation of local authority areas into a single administrative unit outside Greater London. It places Teesside among the most important centres in the country (in Moorsom 1996: 16).

The formation of the County Borough of Teesside gave the wider region an identity of its own. Yet within a very short period of time, plans were made for the complete reorganisation of local government in England and Wales. In 1974, six years after the formation of the County Borough of Teesside, the County of Cleveland came into existence. As the urban centre of Hartlepool was included in the new administrative area, County Cleveland covered a considerably greater area than the County Borough of Teesside had (146,011 acres with a population of 573,500 compared with 49,107 acres with a population of 392,990 respectively) (see Figure 3.7 below).

Figure 3.7: County Cleveland

In 1992 the Local Government Commission for England was established to review the structure of local government in the English shire counties, and it was recommended that Cleveland County Council be dissolved. Four new all-purpose authorities based on the existing Districts within the County were established, and in 1996 Middlesbrough Local Authority came into existence along with Hartlepool, Stockton-on-Tees, and Redcar & Cleveland Local Authorities (see Figure 3.8). 56

Figure 3.8: Four Local Authorities

Each authority is now regarded as a County in its own right. However, ‘for cultural and ceremonial purposes’ the old boundary running along the River Tees has been reinstated and Hartlepool and Stockton-on-Tees have links with County Durham, whilst Middlesbrough and Redcar & Cleveland are associated with Yorkshire (Moorsom 1996: 22). Within a period of approximately 30 years, then, in terms of local government, Middlesbrough has been assigned four separate political identities.

Despite lying on the south bank of the River Tees and therefore having an association with Yorkshire for ‘cultural and ceremonial purposes’ (Moorsom 1996: 22), in economic terms Middlesbrough is the centre of the sub-region of the Tees Valley, which is part of the larger region of the North East. This region can be seen to include the conurbations along the Rivers Tyne, Tees and Wear (where 70% of the population live) and communities in Durham and rural Northumberland. Middlesbrough’s position in the North East is made manifest by its inclusion in the first Regional Economic Strategy ‘One North-East ’ which is a ‘blueprint for a better regional future in the next millennium’ (1999: 4).

In terms of education and research, although the West Yorkshire-based University of Leeds has an outpost in Middlesbrough, the recently formed Arts and Humanities Research Board Centre for North-East England history includes the Middlesbrough-based University of Teesside. The other universities involved 57

in the major new research project which will investigate aspects of the history of the North East are the universities of Durham, Newcastle, Northumbria and Sunderland, all of which are situated north of Middlesbrough.

Crucially, perhaps, in terms of popular culture, the conurbation both north and south of the River Tees has become increasingly associated with the wider area of the North East of England and increasingly dissociated from Yorkshire. Since the independent regional television groupings were formed and began broadcasting in January 1959, Middlesbrough has been included in the independent television group of Tyne Tees Television and not in Yorkshire Television. Similarly, news from Middlesbrough is included in regional newspapers which cover Tyneside, Wearside and Teesside and not in local newspapers such as the Yorkshire Post. Additionally, considerable local rivalry exists between Middlesbrough Football Club, Newcastle United and Sunderland Football Club. In general terms, in the media, Middlesbrough is now referred to as being in the region of England known as ‘the North East’.

From a linguistic point of view, the situation is much the same. In terms of traditional dialectology, the River Tees is often seen as a boundary. Harold Orton, when working on The Linguistic Atlas of England (1978), reportedly insisted that an isogloss be drawn along the River Tees (this being the only river to have an isogloss follow it) because he knew it to be a boundary between Durham and Yorkshire (Clive Upton, personal communication). However, in modern dialect groupings, both Wells (1982) and Trudgill (1990) group Teesside with Tyneside in the ‘north-east’ or ‘far north’. According to Wells, however, the accent of Tyneside differs from ‘typical northern accents’ considerably more than does that of Middlesbrough (typical northern accents being Greater Manchester, West Yorkshire and South Yorkshire, according to Wells (1982: 350)). Although the Tyneside accent is seen as being a great deal more ‘North Eastern’ than the Middlesbrough accent, the two are usually grouped together. Indeed, Trudgill claims:

no one from Middlesbrough would mistake a Tynesider for someone from Middlesbrough - but the accents are sufficiently 58

similar to be grouped together, and sufficiently different from those of other areas. Londoners, for instance, might mistakenly think that Middlesbrough speakers were from Newcastle, but they would be much less likely to think that they were from, say, Sheffield (Trudgill 1990: 77).

Although, impressionistically, the Middlesbrough accent is arguably closer to that of Tyneside than those of Yorkshire, Middlesbrough lies between two regional accent types of BrE which are relatively easily identified by the lay person: that of Geordie, which is the accent of Newcastle and Tyneside, and that of Yorkshire. This also adds to the transitional character of the area and often makes precise identification of the accent by an outsider difficult. The perceived misidentification of the accent, usually as Geordie, is an emotive and revealing point to which we shall return in Chapters 5 and 6.

The history of Middlesbrough as an urban centre is a relatively short one, as noted in section 3.1.1.2. This adds to the lack of a deep-rooted sense of identity in the town. The geographical location of the conurbation around the River Tees, the location of Middlesbrough in relation to the river, and the ever-changing identity of the urban centre all add to its transitional character both geographically and dialectally. The shift in the identity of Middlesbrough, with the recent pulling of the town away from the north of Yorkshire and into the south of the North East, is what is of particular interest to the study, however. This shift is the principal reason for the choice of Middlesbrough as the location for a study in convergent and divergent linguistic trends and their relation with identity and perceptions of identity.

3.1.1.5 The current economic and social climate

Although efforts have been made to diversify the local economy, it still exhibits a high dependence on the manufacturing industry which is vulnerable to trends in the global economy, notably chemicals, steel, engineering, offshore and energy (One North East 1999: 86). The decline in these industries in the second half of the twentieth century, as noted in section 3.1.1.2, has seen the area become an economic black spot. 59

Although unemployment in Middlesbrough has fallen in recent years, the unemployment rate still stands at 8.5% (an 11.5% rate was recorded in 1997). This figure is compared with a 7.2% level within the Tees Valley, a 6.2% level within the North East, and a 3.5% national average (September 2000 figures, Tees Valley TEC 2001). Within Middlesbrough, at ward level, the rate varies between 2.7% and 19.2%. Male unemployment ranges from 3.5% to 27.3%.

The North East, as a region, has been recorded as having the second highest quality of life index nationally by The Reward Group, an independent consultancy producing a UK cost of living report based on criteria including food, fuel, transport and housing costs in relation to disposable income and salaries (www.onestrategynortheast.co.ulc/). However, in the region, educational attainment is lower than the national average, the incidence of disability is higher, and the death rate is higher, with one in five men in the North East predicted to die before they retire (One North East 1999: 21). Crime rates are high, and heroin is cheaper on the streets of Teesside than anywhere else in the UK (Richardson & Maxwell 1998). Indeed, the situation in Middlesbrough is such that the Department for Environment, Transport and the Regions’ Index of Multiple Deprivation, measuring social exclusion in each of the 366 local authorities in the UK, found Middlesbrough to be the worst district in the country in terms of concentration of deprivation (Tees Valley TEC 2001: 11).

Not surprisingly then the population is declining. This decline is both through a falling birth rate and a general migration out of the area. In recent years between 1997 and 2001, the population of Middlesbrough has fallen from 145,400 (Teesside TEC 1997) to 144,100 (Tees Valley TEC 2001). Between 2000 and 2011 the population of Teesside is expected to fall by 10,700 which is equivalent to a 1.6% decline (Tees Valley TEC 2001). Government statistics reveal that in the next 20 years Middlesbrough is predicted to suffer the sharpest population fall in the North East, and the second sharpest fall nationally, after Merseyside. It seems that just as the search for employment was the impetus for the growth in population in the nineteenth century, so it is the driving force behind the 60 declining population in Middlesbrough’s more recent past, its present and its predicted future.

3.1.1.6 The investigation of the researcher’s native variety

A home town locale can be advantageous for a fieldworker in terms of allowing easier access to informants, as well as ensuring knowledge of the variety under investigation. Investigating one’s native dialect is a fairly common practice (cf. Trudgill 1974). In discussing Labov’s (1972) cumulative principle ‘[t]he more we know, the more we can find out’, Trudgill (1983a: 34) notes ‘[t]he more we know about a variety, the more insights we obtain about its nature and structure, and the more we know what questions to ask ourselves next in planning further research’. Having a native speaker’s knowledge of the variety in question is beneficial when assessing where variation may lie and determining which variables to investigate in a study. Added to this, as Trudgill notes (1983a: 43), an interviewer who is a native speaker of the dialect, or ‘less obviously foreign’, may have more success in accessing the informant’s least carefully monitored speech style, as demonstrated by Douglas-Cowie (1978) in Northern Ireland. In her study of Articlave in County Londonderry, Douglas-Cowie, a native of the village, felt that ‘villagers would be more likely to switch to a more standard linguistic code in the presence of a stranger, particularly if he was a well- educated Englishman with an RP or modified regional accent’ (Douglas-Cowie 1978: 39). This, indeed, was found to be the case as, when compared with data collected when she herself was acting as fieldworker, Douglas-Cowie found that ‘the presence of the English outsider very often initiates a switch to more standard speech codes’ (1978: 40).

3.1.2 Dimensions of variation under investigation

After defining the sampling universe and delineating the boundaries of the community of interest in the research, the next step in the design of the study is to determine the dimensions of variation under investigation. To take into account all the dimensions of variation, if indeed this were possible, would mean the sample would be far too large for an analysis of the data to be feasible. 61

Previous studies have revealed problems encountered when aiming to obtain true representativeness of urban areas (see Houck 1968 for an example of a sophisticated sampling procedure designed to obtain representative data which proved to be unmanageable (Milroy 1987b: 16)). Having defined the sampling universe, then, the sampling decisions of the present study are based on judgement, rather than on random selection which seeks to obtain true representativeness of the localities. The design of a stratified judgement sample is based on the principle that ‘the researcher identifies in advance the types of speakers to be studied and then seeks out a quota of speakers who fit the specified categories’ (Milroy 1987b: 26). Two independent social variables are used in the study, and therefore the types of speakers included in the sample are selected on the basis of whether they fit into the categories determined by the interaction of those two variables. The two independent social variables under investigation in the present study are gender and age (due to practicalities, the informants form a socially homogeneous group as noted in section 3.1.3). The importance of these particular extra-linguistic variables in language variation studies has been detailed in Chapter 2.

3.1.3 Design of fieldwork sample

The Teesside study is an apparent time study of language variation. That is, ‘different age groups are observed simultaneously and the observations are extrapolated as temporal’ (Chambers 1995: 193). The study also incorporates a small-scale panel study, as three of the young male speakers are being tracked to assess whether their level of use of innovatory variants increases, decreases or remains stable over real time (see section 2.1.3 for discussion of apparent time and real time approaches). The findings from the panel study will not be presented in the current work, but will appear as a separate future work.

The speakers in the apparent time study are grouped into emically-defined cohorts. As noted in section 2.1.3, an emically-defined cohort groups speakers according to some shared experience of time which can reflect a life stage (adolescence, old age) and/or some aspect of history. This is in contrast with etically-defined cohorts in which speakers are grouped by equal, but arbitrarily 62 determined age spans, for example, decades. Emically-defined age cohorts are used both to represent different life stages and to correspond with the changing identity of Middlesbrough, this being a major focus of the study. To this end, it was decided to include four age cohorts, and to investigate apparent time variation and its connection with the changing attitudes of the speakers and the changing identity of the urban centre.

The four age groups are defined as old (60-80), middle (32-45), young adult (19-22) and adolescent (16—17). When assessing generational differences, the young adult and the adolescent groups, being almost contiguous, can be taken as a combined group of young speakers, thus giving old, middle and young cohorts. Evidence of the three patterns of distribution of variants across age, as illustrated in Figures 3.9, 3.10 and 3.11 (after McMahon 1994: 241), can be sought through an analysis of generational differences.

Figure 3.9: U-curve

increasing proportion of low prestige form

increasing age Figure 3.10: Variant being introduced

higher proportion of variant

increasing age 63

Figure 3.11: Variant being lost

higher proportion of variant

increasing age

Figure 3.9 represents the normal distribution of the U-curve, expected when change is not affecting the linguistic variable. The young and the old speakers characteristically use a higher proportion of low prestige forms, whilst middle- aged speakers tend more towards the high prestige forms. Possible reasons for the higher or lower proportion of low prestige forms depending on life stage are detailed in section 2.1.3. Figures 3.10 and 3.11 represent the patterns of distribution of a variable undergoing change in progress. Figure 3.10 represents a variant being introduced by linguistic change, and thus occurring much more frequently in the speech of the young, whilst Figure 3.11 represents a variant being lost by linguistic change and thus occurring in much higher proportions in the speech of the old.

The division of the 16 young speakers of the sample, who have an overall age range of a mere six years, into two independent age groups, each with a very narrow range, is designed to seek evidence for fine-grained age differentiations in speech. The division is motivated by the fact that the two age groups selected, 16-17 and 19-22, represent two different life stages, adolescence and young adulthood. These life stages centre upon the fact that, in Britain, the age of 18 represents a rite of passage which is probably unequalled in an individual’s lifetime. In the eyes of society, the child becomes an adult at the age of 18, when drinking, voting and marrying all become legal. Behaviour, outlook and lifestyle are all likely to change in this short period of time. Therefore, evidence of fine­ grained differentiations in linguistic behaviour between the two life stages, 64 adolescence and young adulthood, is sought. As noted, adolescence is considered a crucial age for linguistic change, whilst young adulthood is considered a time when, under marketplace pressures, a move towards a higher level of use of high prestige variants is observed. Whether such a difference is in evidence in the speech of the adolescents as compared with the young adults is, therefore, of interest to the study and will be explored further in Chapter 6.

In terms of the emically-defined age cohorts relating to history, the selection of age ranges was made to coincide with the changing identity of Middlesbrough in terms of local administrative boundaries. The oldest age group will have spent the majority of their lives in Middlesbrough when it was situated in the North Riding of Yorkshire. When the county boundaries changed, speakers from this group would have been of an age when their speech was unlikely to be affected by innovations, around the age of the present study’s middle age group. It may be, then, that these speakers demonstrate a lesser use of features associated with the varieties of the North East.

The middle age group encompasses speakers who would have been between the ages of two and 14 in 1968 when the formation of Teesside took Middlesbrough out of North Yorkshire. At the time when the conurbation along the River Tees was brought together and given an identity of its own, the informants from the middle age group would have been children or adolescents. As noted in section 2.1.3, children and adolescents are generally thought to be the innovators of language (Kerswill 1996). Whether any convergence with the North East and divergence from Yorkshire resulting from the changing identity of the region is in evidence in the speech of the middle group compared with the speech of the older group is, therefore, of interest.

The young speakers will have spent their life in the Middlesbrough situated in Cleveland and latterly in Middlesbrough Local Authority. What significance Teesside or Yorkshire has for these speakers will be of interest. In general, then, what effect, if any, the identity of Middlesbrough, both actual and perceived, has on the speech of the different age cohorts is of central interest to the study. 65

Gender differences are also sought in the present study, as noted. This is both because of the highly marked gender differences revealed in the pilot study of Middlesbrough (Llamas 1998), and because of the importance of gender in linguistic variation as outlined in section 2.1.2. With four age groups and a gender balance in the sample, then, practicalities dictated that it was unrealistic to introduce a third independent social variable into the sample design. Thus, the speakers of the sample are not socially stratified. The sample of informants in the Teesside study is taken from a socially homogeneous group classed as ‘working class’ (WC). Given the fluid, symbolic and somewhat arbitrary nature of the classification of social class, it was felt that the speakers themselves were best qualified to assign themselves to a social class. Therefore, the categorisation of speakers as WC is achieved through self-assessment. Informants, when completing biographical data sheets after the interview, were asked to provide information on their ‘assessment of social class’ (see Appendix 1 for an example of a completed Biographical Information sheet2). Only those informants who completed the relevant section of the sheet with the self-assessment of ‘working’ were included in the sample of speakers. The informant’s self-assessment of social class is supplemented with information on occupation, housing and level of educational attainment to ascertain a more detailed picture of the informant’s background which will be of use in any future work which takes class as a social variable.

Likewise, ethnicity is not used as a social variable, and a representative sample of the ethnic make-up of the town was not sought. However, in the event the sample turned out to be fairly representative. Middlesbrough is Teesside’s urban centre with the highest proportion of the population from ethnic minorities (4.4% in Middlesbrough compared with 0.7% in Hartlepool, 0.7% in Redcar & Cleveland, 1.6% in Stockton-on-Tees and 1.5% in Darlington). The largest ethnic group in Teesside is Pakistani. As one speaker of Pakistani origin was included in the sample, which corresponds to 3.1%, the sample can be seen as being

2 Note that the Biographical Information sheet and the rest of the completed interview pack as presented in the appendices of the thesis are not taken from an informant whose speech sample was used in the present study. 66 representative of the ethnic make-up of Middlesbrough. However, this is incidental. A much larger representative sample would be needed to include ethnicity as an independent variable and assess its significance in language variation.

The design of the fieldwork sample is presented in Table 3.1. Eight cells of four speakers are included, giving a sample size of 32.

Table 3.1: Design of fieldwork sample

Young (16-22) Old (60-80) Middle (32-45) Young adult Adolescent (19-22) (16-17) Male Female Male Female Male Female Male Female 4 4 4 4 4 4 4 4

3.2 Method of Data Elicitation

The unique methodological setting which underlies the present study of Teesside English has something of a dual purpose, as noted in Chapter 1. The brief of the scholarship through which the present study has been undertaken was to design and trial a completely new technique of rapid data elicitation which was to become the core of a national collaborative venture. The Survey of Regional English (henceforth SuRE) is to utilise this methodology in a systematic, consistently-collected survey of regional linguistic variation found throughout the British Isles. Although devised with the requirements of a large-scale collaborative survey in mind, the methodology is to be usable in an individual study of social variation of a given area, in its core form or in an expanded form. Hence, the method of data elicitation has also been developed as an original approach for use in the present research. The Teesside study, then, as well as investigating an area hitherto unresearched, utilises and thus systematically trials the innovative data elicitation technique devised specifically for the requirements 67 of the large-scale SuRE project, whilst simultaneously demonstrating how the method can be expanded to fit the requirements of the individual study.

The basic components of the method of data elicitation devised are now presented. This includes both the core method devised for the needs of a collaborative survey of nation-wide linguistic variation, and the additional elements included for use in the study of Teesside English. We begin, in section 3.2.1, by examining the background of the SuRE project, outlining the aims, objectives and difficulties involved in devising a new data elicitation technique for the survey. The basic components of the proposed new method are then detailed in section 3.2.2, with aims, design, trialling, and revisions of the methodology being considered, in terms of both the content and the conduct of the interview. Finally, in section 3.2.3 we consider the additional elements of the method included for use in the Teesside study.

3.2.1 Background to the design of the data elicitation technique

The larger picture of the concept of SuRE has necessarily dictated the design of the data elicitation method presented in this section. Some understanding, therefore, of the requirements of the methodology and the ideas behind the concept of SuRE is necessary for an appreciation of the potential of the new technique.

3.2.1.1 The proposed SuRE project: aims and difficulties

A new national survey of regional variation in spoken BrE was proposed by Clive Upton at the Reading Workshop on Language Variation and Change in April 1997. As the SED (Orton and Dieth 1962-1971), which was carried out in the 1950s, represents the only consistently-collected nation-wide survey of dialectal variation in England, it was felt that a deficiency existed in the knowledge and awareness of current variation in BrE on a national scale.

Many and varied methods of eliciting data for analysis of language variation exist, and research is continually being undertaken in the field. Recent and 68 ongoing projects offer detailed knowledge and insight into linguistic variation and change in Britain (for example, Kerswill & Williams in Milton Keynes 1997, Cheshire, Gillett, Kerswill & Williams in Milton Keynes, Reading and Hull 1999, Docherty, Foulkes, Milroy, Milroy and Walshaw in Tyneside and Derby 1997, Cheshire, Edwards and Whittle’s British Dialect Grammar survey 1989, amongst others). However, researchers wishing to compare their findings with those of another study are faced with individual projects which have different aims and employ different methodologies. This makes direct comparisons of studies problematic (see Foulkes and Docherty 1999a for further discussion). Given the process of dialect levelling (Williams & Kerswill 1999, Watt & Milroy 1999), and the spread of current vernacular changes in certain phonological and grammatical features in BrE, as noted in Chapter 1, the availability of studies which are regionally disparate but directly comparable would be enormously advantageous.

The basic intention of the SuRE project was to create an electronic database of consistently-collected dialectal material from a planned network of British localities which would record and document the facts of linguistic variation throughout Britain. Such a database would permit detailed analyses pertaining to the diffusion of language change and the spread of current vernacular changes in BrE. The form of the survey was to be guided by the necessity for the primary data to be the object of analytical work addressing current research questions concerning levelling and language change by way of diffusion through geographical and social space. At the same time, its form was to be sufficiently broad as not to preclude the potential for analysis which addresses other research questions arising in the future. For this reason the data collection technique was to be unbiased towards any particular research question. An unslanted approach would be adaptable to use by researchers for different purposes. Also, as SuRE is a proposed collaborative project involving researchers with varying interests, a method which can be used as an adjunct to the researcher’s own work is necessary. The technique must be immediately accessible and effective without compromising the fieldworker’s individual research interests. 69

In order for the SuRE project to obtain as complete a picture as possible of regional language variation, data must be obtained which can be analysed on three levels of possible variation: phonological, grammatical and lexical. To discount any of these levels would be to obtain an incomplete picture of regional variation in spoken English found throughout Britain. These multi-levelled data must be comparable across the localities to be studied, permitting quantitative analyses of the different levels of regional and social variation where possible. Additionally, the method must be relatively quick and easy to administer. Researchers must be able to apply the method to their fieldwork with the minimum of prior preparation or administration superfluous to their particular fieldwork needs.

The primary aim of a methodology for the project was to obtain samples of informal speech from which analyses could be made at the phonological level and, to some extent, the grammatical level. As this was a fundamental requirement of a methodology for the project, a problem lay in combining the level of comparable lexical variation with the necessity of obtaining spontaneous speech, as to control the lexical items used in a conversation is to make the interaction less than spontaneous. This control can have the effect of formalising the speech style, thus hindering the possibility of gaining access to the ‘vernacular’ or ‘the speech style in which the minimum attention is given to the monitoring of speech’ (Labov 1972: 208). As the vernacular ‘gives us the most systematic data for our analysis of linguistic structure’ (Labov 1972: 208), it being the speech style least influenced by notions of linguistic correctness (Chambers & Trudgill, 1980: 58), it can be regarded as the style required by the elicitation method of the SuRE project. Identifying and gaining access to the vernacular is problematic, however, as it is the speech style which is liable to disappear under scrutiny. Hence, we are faced with what Labov terms the ‘Observer’s Paradox’ (1972: 209).

This may have implications for the length of time needed for the interview. Douglas-Cowie (1978: 43) found that around an hour of interaction was necessary for informants to ‘settle into’ an interview with an outsider, and so to use speech approximating to everyday interactional style. Prior to this, speech 70 obtained may have been adjusted to a more standard speech code, thus showing different patterns. The effect of the ‘Observer’s Paradox’ may be overstated, however. The skill of the fieldworker in putting informants at ease and in developing a rapport is vital if the informant’s monitoring of speech is to diminish. Being a native speaker of the dialect under investigation or, at least, being ‘less obviously foreign’ is advantageous in this respect, as noted in section 3.1.1.6. The informant’s perceived reason for the interview is also of crucial importance to the level of informality achieved.

Given that the fieldworker endeavours to create as relaxed an atmosphere as is possible in the interview, from a phonological perspective, a relatively short period of time is necessary to obtain enough data with which to make an analysis. Milroy (1987b: 39) suggests that the time needed could be as little as twenty to thirty minutes. Guy (1980: 26) claims that 30 tokens of a given variable is the minimum required to make an analysis which is statistically testable (N=30 being an important dividing line in statistics between large and small samples). For many phonological variables, this can be obtained from less than one hour of speech. Certain phonological variables are less common than others, however, and in some cases a longer sample of speech may be required to obtain the ideal number of 30 tokens. Difficulties apart then, obtaining comparable data which are analysable phonologically is probably the least problematic of the levels considered.

Obtaining comparable data on regional and social grammatical variation is less straightforward and is probably the least explored level of analysis. Included in the grammatical level of analysis are variations in morphology and in syntax. Edwards, Trudgill and Weltens (1984) presented an overview of more than 200 studies of English dialectal variation in The Grammar of English Dialed: A Survey o f Research to extract information on grammatical variation. They found that the vast majority of the information had to be abstracted indirectly through discussions on phonology or lexis wherein grammar played a minor role. Lack of research on grammatical variation also entails a lack of research methodology and data elicitation techniques available to the researcher for an area of linguistic variation which has implications for, amongst other things, education. 71

As with phonological variation, grammatical variation is sought in natural or spontaneous speech. Much greater quantities of speech are required to obtain quantifiable and statistically significant tokens of a grammatical variable, however. This is largely because, unlike phonological variables, the relevant structures are not likely to emerge predictably and regularly. The difficulty of obtaining grammatical data is further compounded by the fact that the interview as a speech event places pragmatic constraints on the emergence of syntactic patterns. Due to the unequal power base between interviewer and interviewee, interrogative constructions are unlikely to be used by the interviewee, but are likely to be plentiful in natural speech. Milroy (1987b: 54) argues that question tags are unlikely to appear in an interview context. Similarly, McCawley (1971: 109) argues that the ‘hot-news’ perfect is inhibited in this context, and Wolfson (1982: 62) suggests that the conversational historic present, associated with a narrative style of speech, is also unlikely to appear. Additionally, in terms of problems encountered in quantifying data, the possibility exists of semantic differences between non-standard and standard grammatical forms being perceived by the informant. Furthermore, the informant may claim knowledge and demonstrate use of more than one non-standard form, and a single non­ standard grammatical form may not be associated with a region or a dialect.

The analysis of lexical variation poses problems similar to those encountered in a phonological or a grammatical analysis in many ways. However, the greatest difficulty lies in developing an effective elicitation technique which combines the level of lexical variation with the other levels of linguistic analysis. To a large extent, phonological and grammatical data can be obtained in the same way, through informal speech. However, informal speech will not guarantee the production of lexical data which are comparable regionally or socially. The fieldworker has no control over the linguistic expressions (here taken to include linguistically simple and linguistically composite expressions) an informant uses in spontaneous speech. To try to control for this drastically alters the desired speech style of the informant, as the spontaneity of the discourse is constrained.

Furthermore, as variants of lexical items involve semantic differences they are not quantifiable, unlike variants of a phonological and, to some extent, a 72 grammatical variable. The choice of [?] as a realisation of (t), for example, does not affect the referential meaning of the linguistic expression in which it is produced. Having no referential distinction, the variant can be analysed and quantified relatively easily. To some extent, grammatical variation can also be analysed and quantified as standard and non-standard forms with the same referential meaning. Lexis, however, is problematic in terms of identifying lexical items as dialectal variants of a standard variable with the same meaning. Without being able to classify variants as synonyms of the variable under question, quantification becomes difficult.

Taking synonymy to be the sense relation of sameness of meaning, it is tempting to consider dialectal variants as synonyms of the standard linguistic expression. However, absolute synonyms are uncommon. It would be unlikely that two words with exactly the same meaning would survive in a language, as Ullman (1957: 108) notes ‘it is almost a truism that total synonymy is an extremely rare occurrence, a luxury which language can ill afford’. Lyons (1995: 61) lists three conditions which must be met for two expressions to be absolutely synonymous:

(i) all their meanings must be identical; (ii) they must be synonyms in all contexts; (iii) they must be semantically equivalent in all dimensions of meaning.

Failure to comply with all of the conditions would result in there being only partial synonymy.

Semantic differences can result from use of different style, register, or expressive or evaluative meaning. The use of dialectal variants of a standard linguistic expression can indicate a different level of formality, hence Lyons’ second condition is not met. The dialectal variant may also not be semantically equivalent to the standard in all dimensions of meaning, as demanded by the third condition. What is being analysed, therefore, is partial synonymy. Partial synonyms are extremely difficult to classify and quantify. Without taking into account the context and connotation involved in the use of particular dialectal variants, their classification into quantifiable categories is fraught with problems. 73

Moreover, the number of lexical variants of a standard variable is unknown to the investigator. The fluidity of vocabulary in the mental lexicon of the individual means that speakers have at their disposal a number of choices for a given linguistic expression. This choice will depend on factors such as register, context, situation and topic. Even if the elicitation of a particular target word is aimed for, other variants may be produced. This makes the process of quantifying the results problematic. Additionally, as the use of a particular lexical item is a choice on the part of the informant, the differentiation between knowledge and use of such items also poses problems for analysis. Notwithstanding the difficulties posed in the elicitation and analysis of phonological and grammatical data, then, the inclusion of a lexical level of variation considerably increases the complexity involved in devising the all-inclusive data elicitation technique which is required for the present study and for the SuRE project.

3.2.1.2 Previous studies and their applicability

As a means of eliciting data, the questionnaire has been employed in traditional dialectology since the nineteenth century and was the ‘fundamental instrument’ of the SED (Orton 1962: 15). (See also use of the questionnaire in The Survey of Anglo-Welsh Dialects (Parry 1977-79), The Tape-recorded Survey o f Hiberno- English Speech (Barry 1981), The Linguistic Survey of Scotland (Mather and Speitel 1975) and the less formal Linguistic Atlas of the Middle and South Atlantic States (LAMSAS) (Kretzschmar, McDavid, Lerud and Johnson 1994).) Although the traditional questionnaire proves successful in eliciting lexical and some grammatical and phonological data, it would be entirely inappropriate for a current survey whose intention is to access and collect samples of informal speech large enough for phonological analyses which permit quantification to be undertaken from them.

Additionally, the methods employed by the SED, and by other studies undertaken within the traditional dialectological paradigm, give scant information on language variation associated with social factors within a given area, this not being the focus of interest of such research, as noted in sections 74

2.1.1 and 3.1.1.1. Social variables, however, are central to current studies of variation. As such, many more informants are required from each location than the two or three used in the SED. Therefore, the method for the SuRE project must be relatively rapid and easy to administer, demanding the minimum of the informant’s and the fieldworker’s time, unlike the lengthy SED questionnaire which contained nine books of questions, each one taking at least two hours to complete (Orton 1962: 17). Thus, methods which are associated with traditional dialectological studies of language variation are quite inappropriate to the proposed SuRE project.

However, methods used to obtain data for research undertaken within a quantitative paradigm also prove inappropriate. The basic techniques of quantitative sociolinguistic methodology were established by Labov in his 1963 Martha’s Vineyard study and his 1966 Lower East Side New York study, and have been adapted and updated ever since. Research undertaken within a quantitative paradigm uses large bodies of systematically collected language data to discover evidence of language variation and change. The data obtained are analysed in terms of a small number of quantifiable linguistic variables from which quantitative statements are made. The variability in speech is shown to be systematic and linguistically patterned. From the evidence of the variability, socially sensitive accounts of language variation and change are developed.

Various attempts have been made to access the vernacular, or the informant’s least overtly careful speech style. A much-quoted technique used by Labov in his Lower East Side study was the use of the question, ‘Have you ever been in a situation where you thought you were serious in danger of being killed—where you thought to yourself, “This is it”?’ (1972: 93). This can produce an involving and emotional narrative, focusing the informant’s mind on the actual content of the story and thus producing a less monitored speech style. However, informants can adopt the rhetoric of narrative in such a context, which is not necessarily how people talk normally. Additionally, the problem exists that, although perhaps producing impressive results in Lower East Side New York, Labov’s question does not yield the same result everywhere. People have not always experienced a 75 life-threatening situation, as Trudgill discovered when he attempted to use the question in Norwich (1974).

Allowing informants to converse in self-selected pairs on topics of their own choosing with minimal fieldworker involvement is a technique which has been used recently (see Docherty et al. 1997, Stuart-Smith 1999). This technique was used in the pilot study of MbE (Llamas 1998). However, although successful in obtaining informal speech, techniques associated with quantitative studies almost completely remove the possibility of obtaining comparable information on lexical variation which must form part of the current method of data elicitation.

The anthropological technique of participant observation has also been used to effect in studies of language variation. Examples can be found in the works of Cheshire (1982) in Reading, who analysed grammatical data in terms of an ‘index of vernacular culture’, the Milroys (1987a) in Belfast, who analysed phonological data in terms of network strengths, and Eckert (2000) in Detroit, who analysed phonological data in terms of speakers’ participation in communities of practice. However, although successful in gaining quantitative and qualitative data, a participant observation technique is far too time- consuming for a large-scale collaborative project. Although the techniques may not be appropriate, the wish to access the vernacular, as in quantitative studies, and the wish to obtain stylistic variation in the speech sample, which is another technique associated with quantitative sociolinguistic methodology, are central to the aims of SuRE.

Thus, because the data must be elicited quickly and easily, and because lexical variation must be included, which in turn eliminates the option of ‘free’ conversation, an interview of some sort must be used in the SuRE methodology. However, a completely different approach to the elicitation of lexical data than that of the traditional questionnaire is necessary, as the interview must elicit data which are analysable phonologically and also grammatically. Consequently, as the data must be quantifiable (where possible), comparable, and analysable on three levels of variation, and as the elicitation technique used to secure the data must be administrable to a relatively large number of informants, it has been 76

necessary to devise a completely new method of data elicitation, as no existing technique has been found to be entirely suitable or applicable to the needs of the proposed SuRE project.

3.2.2 The new method: the SuRE core

3.2.2.1 Overall aims

The primary aims of the new method are to obtain samples of informal speech from the informant (from which multi-levelled analyses of both regionally and socially comparable data are possible), and to elicit the data as rapidly and simply as is possible. A methodology which is perceived to be too complicated or lengthy to administer may result in the unwillingness of potential fieldworkers to use it for their research purposes, and may also discourage informants.

Although the interview as a speech event is not the ideal means through which to elicit casual conversation due to the ‘asymmetrical distribution of power suggested by the roles of questioner and respondent’ (Milroy 1987b: 49) which can have an effect on the speech style elicited, it proves to be the only practical way of obtaining the necessary data. It is vital therefore to lessen the formality of the interview situation as much as possible, and to make the interview an unintimidating and, if possible, enjoyable experience for the informant. This is an issue which is crucial both in finding willing informants and in accessing the informant’s least overtly careful speech style. To a large extent, the informant’s enjoyment of the interview is dependent on the skill of the fieldworker in putting the informant at ease. However, throughout the development and revision of the methodology, ensuring that the informant is comfortable with and interested in the material has been of paramount importance, and a number of revisions have been made by acting on informants’ feedback.

In order to obtain the required informal speech style combined with data on lexical variation in the interview, the fieldworker, using the newly devised methodology, ‘leads’ a conversation around semantic fields. To lessen the formality of the interview context, the interview is undertaken with informants in 77 self-selected pairs, permitting interaction to be more like a conversation than an interview. Discussion on local lexical items is prompted by the fieldworker, with informants encouraged to discuss ‘local’ words, how they are used and what connotations they have. Fieldworker participation in the interaction is minimal, however. Experience has shown that the prompting and encouraging of conversation about local lexical items often results in informants discussing and disagreeing about their perception of dialect words. In itself, this approach is unusual. Dialect surveys such as the SED have avoided posing direct questions to elicit lexical items. Such discussion as does take place concentrates on classifying which items are known and which are used by informants. Discussion about lexical variation which encourages informants to elaborate on attitudes towards their responses is rare. In one of the few recent studies on lexical variation, Macafee (1994: 46) maintains ‘[i]t is difficult to discuss the details of a language or dialect with non-linguists, even with the native speakers’. In terms of lexical items known and used, however, there is arguably no-one more qualified to discuss the details of the dialect than a native speaker, whether a linguist or not. A notable exception to the avoidance of direct questions and discussion about lexical items comes from Pratt (1983: 153), who suggests ‘[o]ne must get the informant to TALK about the word, use it in different contexts, pass judgement on it, in short, to display his knowledge of it.’. As well as demonstrating knowledge of particular lexical items, the prompting and encouraging of discussion about lexical items and lexical variation reveals invaluable attitudinal information about the dialect. Also, a relatively casual conversation centred upon the topic of lexical variation ensures that the attitudinal information and data on lexis are produced within the context of relatively natural speech. As well as producing informal speech from which phonological and, to some extent, grammatical analyses can be made, the ensuing conversation produces a mass of information on the lexical data produced. This can include age and sex differences in usage, connotational and collocational information, knowledge and use differentiation of given items, and attitudinal information on dialect.

Although the method of discussing lexical items in self-selected dyads produces the sample of informal speech for analysis, a conversation alone does not yield 78 quantifiable or comparable data on lexical variation. Control must still be exercised over the specific lexical items elicited in order for direct comparisons of variants to be possible. This should not jeopardise the casual conversational style of the interview, however, so the somewhat formal traditional questionnaire is unsuitable, as noted.

3.2.2.2 Sense Relation Network sheets: visual design and content design

The principal tool devised and designed to allow the information on lexical items to be comparable regionally and socially, and to give a somewhat flexible structure to the interview, is the Sense Relation Network sheet (SRN). The three SRNs which form the core of the interview are shown in Figures 3.12, 3.13 and 3.14 below.

Both the visual design and the content design of the SRNs are inspired by the idea that there exists a ‘web of words’ (Aitchison 1997: 61), or a series of interconnected networks which define, delimit and store linguistic expressions in the mind. The meaning of a linguistic expression is constituted by and is part of a network of meaning which is organised in an intricate, systematic structure in the mind. Within this structure, linguistic expressions from similar semantic fields define and delimit each other’s meaning through their interconnecting sense relations; for example, ‘settee’ defines ‘armchair’ in a way that ‘apple’ does not.

The visual design of the SRNs is also inspired by materials and aids used in language teaching, such as word trees and word field diagrams (see Gairns and Redman 1986). The importance of visual impact in language teaching materials is well known (Gairns and Redman, 1986: 96; Ur, 1988: 20). Visual impact engages the learner and assists in the retention of information. Likewise, to attract the attention and engage the interest of the informant, the visual impact of the SRNs is crucial. 79

Figure 3.12: Sense Relation Network sheet (first of three)

intelligent stupid soft shoes (worn by children for P.E.)

grandmother / father 80

Figure 3.13: Sense Relation Network sheet (second of three)

pregnant mad drunk 81

Figure 3.14: Sense Relation Network sheet (third of three) 82

As can be seen in Figures 3.12, 3.13 and 3.14, visually, networks are designed in which the standard notion words are connected to subdivisions. The subdivisions, in turn, are connected to the semantic field of the SRN, symbolising, in a way, the interconnected network or ‘web of words’. Space is then provided under the standard notion word for the insertion of a dialectal partial synonym. Each SRN is printed in a different colour, the aim being for the visual impact of the SRNs to be positive and unthreatening, and for the SRNs to engage the interest of the informant to a level at which the desire is to complete them.

In terms of content design, the SRNs are built around semantic fields (Lehrer 1974) and, as such, are akin to the grouping of questions by subject matter in the SED questionnaire. According to Johnston (1985a: 83), the grouping of questions by subject matter, as opposed to alphabetically or randomly, allows for a level of spontaneity in the responses. On the SRNs, standard notion words are offered as prompts for the elicitation of dialectal variants, as interviews which use indirect elicitation techniques are much more time-consuming than those which use direct ones. Additionally, indirect questioning may make the interaction feel more like an interview or a test of knowledge than a conversation, so skewing speech style towards the formal.

The selection of semantic fields and standard notion words in the three SRNs is the result of trialling and revision of the method during which eight original SRNs have been subsumed under the present three3. The subsumption was made in the interests of reducing the time needed by informants to complete the SRNs, as well as reducing the time necessary to conduct the interview. None of the initial semantic fields have been discarded entirely, but the fields have become broader to encompass a greater area of notion words. Standard notion words producing little or no variation in trialling were removed. However, each sub­ division carries space for dialectal variants of notion words not on the SRN which the informant may wish to include. When selecting the original semantic fields, sub-divisions and notion words, reference to groupings and lexical items

3 The original eight semantic fields were; Qualities and Traits, Relationships, States and Feelings, Physical Appearance, Actions, Daily Life, Speech Acts and Physical World. 83 found in previous questionnaires were used (for example, SED, LAMSAS) as well as introspection. Reference also to books on specific dialects was made, such as A Century of Yorkshire Dialect (Kellett & Dewhirst 1997), Geordie words and phrases (Todd 1987). Additionally, the wish to include the same standard notion word as the SED where possible and appropriate was borne in mind, as a direct comparison could reveal possible real time change. Due to the urban bias of the proposed survey and of the study of Teesside English, however, this proved inappropriate in most cases, and few SED notion words were retained.

The SRNs then, as well as being a visual network, rather than a list of questions, represent the interrelated network of paradigmatic and syntagmatic sense relations in which linguistic expressions from similar semantic fields define and delimit each other’s meaning. They also represent the sense relation of partial synonymy, which the dialectal variant holds with the standard notion word. Additionally, in time they may represent a geographical sense relation network of dialectal variation of partial synonyms found throughout the British Isles.

3.2.2.3 Sense Relation Network sheets: technique of administration

Coupled with their concept and design, the technique of administering the SRNs is an essential part of their success as a method of eliciting data. Informants are given the SRNs some five days before the interview, with both verbal instructions from the fieldworker and written instructions as part of the interview pack (see Appendix 2 for instruction sheet4). The innovatory step of allowing informants to know the content of the interview prior to the event has implications both for the content of the interview and for the interview as a speech event.

4 Note that the instruction sheet shown in Appendix 2 is part of an interview pack used by an informant from the Teesside study (not an informant who was included in the final sample, however). As such, this carries an additional instruction about the completion of the Language Questionnaire, which is part of the extended methodology used in Teesside (see section 3.2.3.1). This, and therefore the instruction, do not form part of the core methodology. 84

Giving informants the SRNs prior to the interview allows them time to consider the lexical items they use. This has a dramatic effect upon the amount of lexical data yielded from the interview. If they are asked to produce a dialectal variant as an immediate response to a prompt, there is a danger of the informant’s mind going blank. This results in minimal data being yielded. This may also necessitate an undesirable level of prompting from the fieldworker. More importantly, however, there could be a harmful effect on the required speech style and the willingness of informants to speak at length, due to a feeling of unease in the interview situation. Thus, the technique of administering the materials prior to the interview maximises the amount of data yielded. This can be seen through the results of an experiment which was carried out with a group of 17 year old ‘A’ level English language students from various parts of the British Isles to ascertain how much lexical data would be yielded through eliciting immediate responses to the prompts. The students were given the three SRNs and approximately 15 minutes to complete them. On average, without allowing the students time to consider the dialect words they use and without having the benefit of discussion, an average of 53.8 lexical variants were produced through use of the three SRNs. This is in stark contrast to the average of 111.3 lexical variants per speaker recorded in written form on the SRNs from the present study. Therefore, any perceived benefits to obtaining an initial response to the prompt word cannot outweigh the potentially huge difference in data yielded through giving informants time to consider and discuss the dialect words they know and/or use.

In terms of the interview as a speech event, any feelings of unease in the interview situation might be heightened if the informant perceives the interview as a test of some sort. By having prior knowledge of the content of the interview, however, it is thought that suspicion on the part of the informant is diminished considerably. This, combined with the fact of experiencing the interview in self­ selected pairs, allows informants to settle into a relatively casual speech style in as short a time as possible. To ensure the ready recruiting of informants and to maximise the possibility of gaining access to their least overtly careful or monitored speech style, it is crucial that informants feel at ease and enjoy the interview as much as possible. 85

When the informants have had some days in which to complete the SRNs at their convenience, discussing responses with others should they wish (differentiating between their own and others’ responses on the SRNs), the paired interview is undertaken and recorded. The interview consists of the written responses on the SRNs being read out by the informants, and responses being discussed in terms of whether informants use the variants or only know them, situations in which they would be used, connotations and collocations associated with the variants, and anything else which informants may initiate. The fieldworker can use an interviewer’s guide to ensure that all the notion words are covered (the informants keep their own SRNs until the end of the interview). The interviewer’s guide can also contain prompt questions, for example, the use of intensifiers, gender differences in use, age differences in use, varying degrees of a state (for example, very cold, a little cold), additional notion words or senses of the notion words given, all of which can provide additional information and extend the discussion. During the interview other known or used variants which are produced are noted on the SRNs in different coloured ink by the informant. Thus the written record of the informant’s responses on the SRNs (which the fieldworker collects after the interview), a recording of the informant’s spoken responses for pronunciation purposes, and a mass of attitudinal information on the lexical items elicited in an informal speech style are all secured by means of the SRNs and the recorded interview.

3.2.2A Sense Relation Network sheets: trialling and revision

The trialling of the SRNs was undertaken to ascertain how effective the method was in eliciting data for the multi-levelled analysis and how easy the technique was to administer. The data obtained through trialling were not to be analysed to make quantitative or qualitative statements about linguistic variation. Consequently, the informants were chosen by virtue of their being willing and able to participate, rather than by sophisticated sampling procedures. Coincidentally, the informants constituted a fairly even gender balance and covered a wide range of ages (from 14 to 71 years of age). Also, the informants were all from the same socio-economic background, broadly speaking WC. 86

Four trials of the method were undertaken with 12 informants from Leeds. From the outset the method proved to be successful in eliciting casual conversation and in yielding a substantial amount of information on lexical variation. Additionally, all informants expressed an enjoyment of the interview. The revisions to the technique were made chiefly at the level of the individual standard notion words. Notion words not revealing dialectal variation were removed. This led to a broadening of the sub-divisions and, in turn, the semantic fields of the SRNs. After the second trial the eight SRNs were subsumed under the present three, as noted. Thus a beneficial reduction in the time needed for the interview and the time necessary for the informants to complete them was achieved. As well as being trialled with 12 informants from Leeds, the technique has been tried by other researchers in an external trialling stage of its refinement.5 Apart from the reduction in the number of standard notion words and the consequent subsumption of the original eight SRNs into three, no major changes were made to the technique before its use in the Teesside study.

3.2.2.5 Sense Relation Network sheets: data yielded

In terms of lexical items elicited through the SRNs, the richness of the data yielded can be seen in the three completed SRNs which appear as Appendix 3. The potential for the study of the differences and problematic distinctions between dialectal variants, regional slang, national slang and standard colloquialisms is clear. The study of non-standard orthography is also promoted by the method. Additionally, the difference between items produced before and items produced during the interview may be of interest.

5 Thanks go to Ann Williams, Jason Jones, Mark Jones and Louise Mullany for trying the method and giving helpful comments on the effectiveness of the technique as a method of data elicitation. The method has also been used by students from the University of Leeds and the University of Basel. 87

More lexical data are produced from the recorded discussion about the responses.6 Informants can use dialectal variants without necessarily being aware they are doing so. For example, one informant, when discussing the notion word ‘man’, claimed that she would never use bloke after already having done so during the interview. Additionally, informants may become aware only when they hear someone else use it that they themselves use a particular word. Also informants’ insights into which variants are considered to be local, as opposed to those which are more widely used, can be revealed. For example, one informant claimed not to have inserted a variant for soft shoes worn by children for P.E. because she ‘couldn’t think of another word for sandshoes ’, indicating that she believed sandshoes to be a widely used or standard variant.

Once read in isolation, lexical items are immediately put into context by the informant. Thus, the individual lexical item is clearly recorded for transcription purposes and can then be disregarded for the purposes of a phonological analysis of informal speech (the nature of the written response on the SRN being read aloud possibly constituting a more formal reading style of speech). It would, however, be possible and interesting to compare phonological features of the more formal and less formal styles. The context of the interaction makes it clear which particular lexical items are read aloud and which are not. Alternatively, the use of different coloured ink on the SRNs is an indicator of which variants were written before the interview (and thus read aloud), and which were noted down during the interview (written after having been spoken).

After having been read aloud, the lexical items are generally elaborated upon and discussed in the context of casual conversation, giving the sample of informal speech which can be analysed phono logically and grammatically. For example, after having given the responses twoc, tax, nick, skank, and swipe for the notion word ‘steal’, two informants went on to discuss at length precisely what each term referred to and their ideas on the origins of the words. Similarly, gender and

6 Although the three SRNs shown as Appendix 3 present 215 variants for 80 standard notion words, by including all the variants the informant mentioned but did not write on the SRNs during the interview and those she claimed knowledge of during the recorded interview, a total of 272 variants were counted from this informant. 88 age differences in responses to notion words are discussed at length, with, for example, two young male informants arguing that they would never use the variant bonny for the notion word ‘attractive’, it being an ‘old person’s’ word, and they would never use canny-looking, it being used by females, opting themselves to use nectar, sweet, fit and lush. Thus, the informal speech which can be analysed phonologically and grammatically also contains a mass of data on knowledge and use of lexical items, attitudinal information on dialectal variants, ideas on word origins, changing societal attitudes to lexical items and perceptions of and actual gender and age variation in usage. In this way a multi­ levelled bank of data is produced through use of the SRNs.

3.2.2.6 Identity Questionnaire

Combined with the three SRNs, an Identity Questionnaire (IdQ) is included in the interview. The IdQ is given to the informants, with the three SRNs, prior to the interview, these forming the interview pack. The questions posed in the IdQ of the core interview for the SuRE project are listed below in Figure 3.15. The IdQ can be expanded for use in a given area as in the Teesside example taken from the interview pack (see Appendix 4).

The core IdQ comprises 15 questions whose aim is to obtain an insight into informants’ attitudes towards their language and their area. The information obtained can be correlated with linguistic variation found. In this way the linguistic data can be analysed within a language ideology framework as detailed in Chapter 2. As communities and boundaries are often symbolic, it is problematic to impose a definition of speech community onto a geographical area and a group of people, as noted in section 2.1.1, even when an investigator is a native of the geographical area to be studied. The similarities and differences which define and delimit communities are often not a matter for objective assessment, but are largely subjective, existing in the minds of the members of the community (Cohen 1985: 21). There is no reason why the topical content of the interview should not be of use in this regard, with the fieldworker tapping the natural resource of the informant for information on language, area, boundaries and attitudes. 89

Figure 3.15: Identity Questionnaire

Your Language

• What accent would you say you had, and do you like it?

• Can you recognise the accent of your home town (e.g. if heard on the radio or

T.V.)? If so, how?

• Do you think older and younger people talk the same here (pronounce things the same and use the same words)?

• Have you ever been in a situation where you’ve deliberately changed the way you

talk? If so, why?

• Do you think there’s a difference between how males and females speak here?

• Where, geographically, would you say people stop talking the same as you and

start sounding different?

Your Area

• If you were watching a regional news programme, what places would you expect

to hear news from?

• What image or description of your home town would you give to someone who didn’t know it?

• If you wanted a day out shopping, where would you go?

• What do you consider the local football derby to be?

• If you could, would you change where you came from? Why/why not?

• What do you consider the best and worst things are about growing up and living in your home town?

• Have you ever seen your home town on a national T.V. programme (e.g. a

documentary)? If so, how was it portrayed?

• If an outsider was complaining about your home town, would you defend it even if you agreed with what s/he was saying? Why/why not?

• How many friends, relations and work/school/college mates do you have in the

neighbourhood (not more than about 5 mins. away) who you se6 regularly? 90

In the IdQ, questions are posed to elicit information about people’s attitudes towards language and identity (Le Page & Tabouret-Keller 1985), existence and awareness of age and sex differences (Kerswill 1996; Kerswill and Williams 1997; Milroy, Milroy and Hartley 1994; Trudgill 1974), and rudimentary ideas on density of networks (Milroy 1987a). They may also elicit information on people’s perception of language areas and boundaries (Preston 1989) and information on awareness of and reasons for speech accommodation (Giles and Powesland 1975). In this way a mass of attitudinal information is gained from the individual through the use of the IdQ. This attitudinal information can provide comparable data across regions of Britain and may reveal differing regional attitudes towards areas and dialects, as well as revealing possible gender, age and class variation in a given area.

This then is the core method of the proposed survey designed to elicit a sample of informal speech.7 From the speech sample data are obtained for a multi­ levelled analysis of phonological, grammatical and lexical variation. A word list may be included in order that stylistic variation can be observed. Because of the larger amount of data necessary for grammatical analyses, and the structural limitations of the interview which places ‘pragmatic and discourse constraints on syntactic structure’ (Milroy 1987b: 56), it is anticipated that a more formal grammatical element will be included in the methodology for the SuRE project.

7 In order to test the level of formality which is achieved in the interview situation as opposed to recorded ‘free’ conversation, data have been obtained from six informants in both ‘free’ conversation style and interview style. The ‘free’ conversation was obtained by allowing the informants to chat in social pairs at their own convenience without the fieldworker present (Llamas 1998). The use of certain localised variants was then compared to ascertain whether a dramatic shift to a more formal speech style was indicated in the interview situation. No such shift was revealed. Additionally, all six informants, when asked, claimed to have preferred participating in the interview using the methodology presented in this phapter rather than recording themselves in ‘free’ conversation. 91

3.2.3 The new method: additions for the Teesside study

As noted, the core methodology can be used in a study of sociolinguistic variation of a given area. This can be either in its core form or in an expanded form. For use in the Teesside study several additional elements have been included, as outlined below.

3.2.3.1 Language Questionnaire

In order to obtain awareness of informants’ perceptions of the non-standard grammatical features found in the area, a Language Questionnaire is included in the Teesside interview pack (see Appendix 5 for a completed Language Questionnaire). Informants are given a list of 40 sentences, each containing a non-standard grammatical construction or a regional discourse marker. Informants mark which features they would hear in the area, use themselves in speech, or use in written form. The responses to the questionnaire can then be compared to and correlated with informants’ actual usage of non-standard grammatical features in informal speech.

The Language Questionnaire is based on the type of questionnaire used by Cheshire et al. (1989) for the Survey of British Dialect Grammar. The majority of the sentences used in the Language Questionnaire are authentic, having been taken from the recordings of ‘free’ conversation for the pilot study of Middlesbrough (Llamas 1998). Many grammatical features included in the questionnaire are features associated with urban varieties of BrE, for example, multiple negation, them as demonstrative adjective, what as subject relative pronoun, present participle sat, never as past tense negator. Given the geographical position of Teesside, certain sentences have been included to ascertain whether features associated with a variety from further north, namely Northumberland English, are used. Questions 35, 36 and 37 have been taken from Beal (1993). Other sentences contain features associated with Yorkshire English, for example, questions 10, 15, 17. The findings from the Language Questionnaire will appear in future works which investigate the use of non­ standard grammatical forms in the data. 92

Along with the Language Questionnaire, which constitutes the more formal grammatical element, a word list is included in the Teesside study to facilitate observation of stylistic variation and for control of environment in the phonological variables analysed. The citation forms used in the reading list are presented in Appendix 7 and have been selected largely in order to be directly comparable with the ones used in the pilot study (1998) which, in turn, were intended to be directly comparable with those used in the study of Tyneside English by Docherty et al. (1997).

3.2.3.2 Teesside Identity Questionnaire

The IdQ used in the Teesside study contains four additional questions (these can be seen in Appendix 4, questions 7, 9, 10 and 13). Due to the Teesside study’s focus on the transitional nature of the area, its changing local identity, and convergent and divergent linguistic trends as noted in section 3.1.1.4, questions are asked which aim to elicit responses about feelings towards the changes in the county boundaries and the location of the area. Also, the informants are asked about their feelings on a perceived misidentification as Geordie or Yorkshire.

The responses to the IdQ will give an indication of how closely tied to their area and proud of their accent and dialect the informants claim to feel. The informant’s level of identification with the area is of importance to the study. The responses to the questions on the IdQ will give an insight into whether informants feel positively, negatively or neutrally towards their dialect and area. Responses to certain questions can be judged to be ‘positive’, in that they express positive feelings towards the area and the dialect; ‘negative’, in that the informant expressed a desire to live elsewhere and a dislike of the dialect; or ‘neutral’, in that the informant’s feelings seemed neither positive nor negative, or perhaps a combination of the two, much like a section of Labov’s (1972: 39) work in his Martha’s Vineyard study. Not all of the responses to the questions in the IdQ will lend themselves to a classification of this kind (for example, density of social networks, perceptions of boundaries), but many will (for example, whether or not the informants like their accent, whether they would change where they come from if they could). A subjective judgement, therefore, can be 93 made by the researcher as to whether the informant responded positively, negatively or neutrally to their language and their area in the IdQ as a whole.

3.2.3.3 Affiliation Score Index

To counteract the subjectivity of this decision somewhat, a more objective Affiliation Score Index has been devised (see Appendix 6). This is an adapted and extended version of the Identity Score Index used by Underwood (1988) in his study of Texan English. This was an attempt by Underwood to use Le Page and Tabouret-Keller’s (1985) theory of acts of identity to account for linguistic variation in Texas. Le Page and Tabouret-Keller had intended the theory of acts of identity to be ‘universally applicable’ (1985: 182), as noted in section 2.1.1, and it had previously been used by Trudgill (1983b) to account for the variation in British pop song pronunciations. Underwood (1988: 410) constructed an ‘Index of Texan Identification’ with which he scored responses to three questions designed to test the level of local affiliation. When analysed, the use of the localised variant under consideration was found to have no linear relationship with social variables, the localised variant appearing dominant in all groupings. There was, however, a clear linear relationship between scores on the index and the use of the localised variant. That is, the closer the informant identified with the group in question, Texans, the higher the use of the localised variant.

This idea has been used in the present study of Teesside English. The Affiliation Score Index comprises seven questions designed to test how closely or how loosely tied to the area the informants feel. The Affiliation Score Index includes direct questions about how the informants feel towards other people from their home town, and questions which test in-group preference. The Affiliation Score Index is not designed to elicit any linguistic data, but simply comprises seven multiple choice questions. The Score Index is administered when obtaining biographical data from the informant. The questions are short and an immediate response is sought. Therefore the Affiliation Score Index is included in the brief final section of the interview in which informants note down their personal details on the biographical information sheet. 94

Each of the multiple choice responses given in answer to the questions carries a score of 1, 2, or 3. A score of 3 indicates the strongest feeling of local affiliation. The scores are calculated and categorised as points along a cline from negative to positive. Broadly, scores are grouped into two categories: negative to neutral and neutral to positive. Linguistic variables and extra-linguistic variables under consideration can be correlated with the results. The levels of local affiliation expressed by the informants through the IdQ and the Affiliation Score Index will be considered in Chapters 5 and 6.

3.3 Fieldwork

With the sampling universe defined, with boundaries roughly delineated, the dimensions of variation under investigation decided, the sample size fixed, and the method of data elicitation designed, the fieldwork was begun. The fieldwork was undertaken between November 1998 and November 1999. Although the findings are based on data taken from the speech of 32 informants from Middlesbrough, interviews were actually undertaken with 75 informants during that time. Various reasons can be offered for the unsuitability of certain speakers. The study was initially designed to include an urban centre north of the River, namely Stockton-on-Tees. This is due to the fact that the Tees is considered a barrier for some variables, for example, ‘h’ dropping south of the Tees (Milroy 1987b: 114), and the fact that the differing histories and identities of the two urban centres are of interest to the study. As this was felt to be an unmanageable task for one person within the time constraints, all data from Stockton-based informants are left for a future comparative work. Also, informants who completed their biographical data sheets with a social class self-assessment of ‘lower middle’ or ‘upper working/lower middle’ were omitted from the sample. Again, a study using social class as a variable is planned as a future work. Another reason for the omission of speakers from the sample was age-based. As the interviews were undertaken with self-selected pairs, on occasion an informant’s chosen partner fell outside the age brackets selected in the design of the fieldwork sample. As all speakers appeared keen to participate and the informant’s enjoyment of the interview was paramount, it was felt inappropriate to deny the participation of a speaker’s partner on the grounds of their age and 95 ask for another partner to be selected. Also interviews were discarded if the recording was not of sufficiently high quality, due usually to outside noise or the presence of small children in the room. Aspects of all the data in the corpus will be used in future works, but, for the moment, all further comment is restricted to the 32 informants whose speech was analysed for the present study.

As all interviews except two were undertaken in self-selected dyads, such as friends, siblings, partners, the informants could not be selected randomly (two interviews were in a one-to-one format as the interviewee’s partner had been unable to attend at the last minute). Therefore, informants were taken from pre­ existing social groups. Although not forming social networks themselves, informants were contacted through a social network approach (cf Milroy 1987a). Such an approach is thought to allow the investigator to obtain larger amounts of spontaneous speech. As such, the study ‘may be capable of describing language variation in greater depth, tapping dimensions of variation which are not obtainable by a survey which samples isolated individuals across a whole city’ (Milroy 1987b: 36).

Although being a native of the area under investigation can ensure relatively easy access to informants, only six of the 32 informants were known to me personally. The other 26 were contacted through a ‘friend of a friend’, which defines the fieldworker’s role somewhat differently from that of researcher (see Milroy 1987a for further discussion). Individuals with connections in social clubs and societies in Middlesbrough were sought (youth clubs, local history societies, writing clubs). This led either to my introduction to the group, as in the case of the local history societies whose meetings I attended and thus met potential informants, or my introduction to individuals through a key contact. Most people who participated in the interview expressed their enjoyment of it and many went on to recommend other people to me who they thought would also like to participate in the study. Approximately half of the informants who were not known to me personally were contacted in this way. No one was paid for their participation in the study. 96

A strict selection procedure was not employed regarding informants’ precise geographical origin within Middlesbrough. A conscious effort was made not to concentrate on any one ward of Middlesbrough, but to include informants from a wide geographical spread. All informants except one had been bom in Middlesbrough (the informant who was born outside the area had come to Middlesbrough at the age of two). In order to ascertain some measure of the ‘localness’ of the informant, information was sought on the birthplace of both parents and grandparents on the biographical data sheets. Additionally, information on whether the informant had lived outside the area and for how long was sought in order to assess the level of geographical mobility of the informants. Information on the birthplaces of the informants’ parents and grandparents and the extent of their mobility is given in Appendix 8. Only mobility outside Teesside is noted in the ‘localness’ charts. Additionally, indication is given as to how recent the experience of mobility was in the case of the young speakers.

After initial contact was made, informants were visited or contacted by telephone with details of the interview procedure. Interview packs were then given or sent with instructions for completion, before a mutually convenient time for the interview was set up some days later. The self-selected dyads in which the informants were interviewed were both mixed-sex and single-sex. The distribution of mixed-sex and single-sex dyads was approximately equal across the sample, with 11 mixed-sex dyads included and 10 single-sex dyads recorded. The possibility of the pairing as a mixed-sex dyad or a single-sex dyad affecting the kind of language used is acknowledged, and may be investigated in a future work. More attention was paid to the avoidance of the pairing of informants across generations, as it was felt that certain questions in the IdQ would generate more discussion if the informants were of a similar age (for example, questions 9, 13). The pairings, then, were either in a friend to friend format (usually same sex) (15 informants), husband/wife format (eight informants), or with a sibling (two informants). One mother and son pairing was included, and one informant was interviewed with a neighbour. The average length of the interviews was 94 minutes. All interviews were over an hour in length and most lasted between 90 minutes and two hours. In total over 36 hours of material were recorded and 97 analysed for the study. The interviews were recorded onto a Sony MZ-R30 Portable MiniDisc Recorder with a stereo unidirectional ECM-DS70P Electret Condenser Microphone.

We turn in the next chapter to the findings of the study and an analysis of the linguistic variables under investigation. Chapter Four Results: Linguistic Data

4.0 Introduction

In this chapter, the linguistic variables under investigation are presented. The variables in the study have been chosen to assess how far, if at all, certain of the current vernacular changes of BrE have been introduced into MbE, as noted in Chapter 1. Additionally, whether any convergence of MbE with varieties further north in the North East of England is in evidence is relevant to the selection of linguistic variables. Only consonantal variables are examined in the present study. The analysis of vocalic variables is reserved for a future work. The present study offers a more extensive and detailed examination of several variables which were examined in the pilot study of MbE undertaken in 1998. An auditory analysis of the following linguistic variables has been made:

(t) [ht] - [f ] ~ [?] ~ 0 (pre-pausal and turn-final)

(p) [p]~~[5p] -[?] (word-medial)

(t) [t]~ (word-medial)

(k) [k] -~t?k] ~[?] (word-medial)

(r) [r]~ M (word-initial & -medial)

(th) [0 ]'~[fj (word-initial, -medial & -final)

(dh) [6]-“[V] (word-medial & -final)

Scores for each variant are recorded as a percentage of the total number of occurrences of the variable. The results from the individual speakers of the cell are conflated, and the group averages in percentage form are presented. Scores for each variant of the variable under consideration are kept separate in the presentation of the results rather than use being made of an aggregate score for the variables in the form of an index score. In keeping scores for each variant 99 separate, the problem of ordering variants in a scale is avoided and a more accurate picture of the variation is revealed (Romaine 1980). One-tailed /-tests were run on the data to test for statistical significance. However, in many cases this appeared unnecessary, and the results of the statistical analysis are presented only to a limited extent

The presentation of the results for each variable will begin with a description of the variants under investigation. This will include details of their social and geographical distribution in BrE. Particulars of the data analysed for the present study will then be outlined briefly. The presentation of the data for each variable will then proceed with details given of the overall distribution of the variants under investigation. This will be followed by discussion of generational differences in the findings, as the data from the combined group of young speakers are compared with those from the middle and old speaker groups. Gender-correlated variation in the data will next be presented. This is followed by the interaction of age and gender, with any fine-grained age differentiation revealed in the data as results from the two young age groups are analysed separately. Data from citation forms will then be presented, again beginning with an overall distribution of variants, before the data are analysed by age and gender.

We begin with findings for pre-pausal and turn-final (t) in section 4.1. Glottalling and glottalisation of (p t k) follows in 4.2. Section 4.3 reveals the distribution of variants of (r), and section 4.4 presents the distribution of variants of (th) and (dh). Finally, a summary of the linguistic trends uncovered in the data will be offered in section 4.5. 100

4.1 Pre-pausal and turn-final (t)

4.1.1 Variants under consideration

Five variants of (t) are considered in two discoursal positions. The variants are

the pre-aspirated alveolar voiceless stop, [ht], the alveolar voiceless stop, [t]1, the

unreleased alveolar voiceless stop, [tn], the glottal stop, [?], and the zero

realisation, 0 . The two discoursal positions are pre-pausal and turn-final. Of the five variants under consideration, previous phonological and sociolinguistic accounts have tended to concentrate on the distribution of released [t] and glottal

[?] forms of (t). Less is known about the geographical and social distribution of the other three variants, [ht], [f ] and 0 .

Ladefoged and Maddieson (1996: 70) describe [ht] thus: ‘[i]n pre-aspirated stops there is a period of voicelessness at the end of the vowel, nasal, or liquid preceding the onset of the stop closure’. Perhaps the best-known examples of pre-aspirated stops are found in Icelandic and Scottish Gaelic, where they are found to occur word-medially and word-finally, with the aspirated stops ... 9 occurring word-initially. In Newcastle English, Foulkes, Docherty & Watt (1999: 12) describe the acoustic characteristics of the phenomenon as ‘a period of high-frequency frication before the voiceless stop gap, or as a breathy continuation of a preceding vowel’. The term ‘extended frication’ is also used by Docherty & Foulkes (1999) and Allen (2001) to describe the acoustic characteristics of a phenomenon which entails ‘a period of fricative energy preceding the stop gap, presumably created by a relatively slow tongue tip closure’ (Docherty & Foulkes 1999: 62).

1 In the present study, aspirated or spirantised stops are included in the category [t], 2 Studies of Icelandic have used [ht] rather than [ht] to indicate pre-aspiration (see Thrainsson (1978)) as ‘pre-aspiration is longer than the aspiration after a stop release’ (Ladefoged and Maddieson (1996: 70). The phenomenon is indicated by the diacritic [h] in the present study, however, given the auditory nature of the analysis. 101

An unreleased stop, [f], one in which ‘[t]he closure and compression stages are made, but the lung pressure then ceases and we go into normal breathing without any explosion’ (O’Connor 1973: 134), can also be found in final position in an utterance. Cruttenden (1994: 145) describes the unreleased variant thus: ‘[i]n final position (i.e. before a pause) ... the closure stage may be maintained, the air compression becoming weak and the release being achieved by a gentle, delayed, and relatively inaudible opening of the oral closure’. Fabricius (2000: 85) describes the identification of [f] as ‘a transition to alveolar position, followed by a hold stage, without the aspirated release of the [th] variant’.

4.1.2 Geographical and social distribution of variants under consideration

Of the five variants under consideration, previous sociolinguistic studies have tended to focus on the distribution of [t] and [?], as noted. Little evidence exists for the distribution, both social and geographical, of the three other variants under consideration, [ht], [f] and 0 .

Within Britain, as mentioned, [ht] is found in Scottish Gaelic (Oftedal 1975:

132). However, evidence for the existence of [ht] in BrE is sparse. The only account of sociolinguistic patterning of [ht] in a BrE variety, to my knowledge, is found in the studies of Newcastle English (Foulkes, Docherty and Watt 1999, Docherty and Foulkes 1999, Allen 2001). Both age- and gender-correlated variation are in evidence in the findings of Foulkes, Docherty and Watt (1999) and Docherty and Foulkes (1999). Pre-aspirated variants are found to be significantly associated with WC female speech (63% of tokens are pre-aspirated in young WC female speech compared with 13% for males) (Foulkes, Docherty & Watt 1999: 14). Furthermore, with data drawn from acoustic analyses of tokens extracted from word-list readings, a comparison of the distribution of variants of pre-pausal (t) in Newcastle English and Derby English revealed that tokens with extended frication occur in data taken from the Newcastle corpus only (Docherty and Foulkes 1999). In Newcastle English, then, the data are found to be ‘suggestive of a change in progress, in that younger speakers are 102 employing a different articulatory strategy for pre-pausal [t], with females leading the way’ (Docherty and Foulkes 1999: 66). In Hull, although not mentioned as a feature of (t), it is reported that final (k) may be pre-aspirated (Williams and Kerswill 1999: 147), as it also is in Highland English (Shuken 1984). Although it may be more widespread than is thought, the evidence to date would suggest that pre-aspirated (t) is an innovation of the North East of England.

The fully released [t] is the non-localised pronunciation found in BrE. However, in pre-pausal and turn-final position, the released stop can also be considered localised to North Eastern varieties. Glottalisation of (t) is a salient feature of both Newcastle and Durham English (Wells 1982, Kerswill 1987, Watt and Milroy 1999). However, constraints can be seen to be placed on the realisation of (t) due to its place in the utterance. Neither glottal nor glottalised realisations of (t) occur in the vast majority of pre-pausal and turn-final tokens. Kerswill (1987: 47) notes ‘[m]y data suggests that, unlike in much RP, word-final, pre-pausal /t/ is never glottalised, but is always a released, heavily aspirated [th].’ Similarly,

Docherty and Foulkes (1999: 62) note ‘[ajlthough glottal variants are widespread in various phonological contexts in Newcastle, they are almost categorically prohibited in pre-pausal position. Tokens before a pause are instead — from an auditory perspective - clearly ‘released’ voiceless alveolar tokens’. Local, Kelly & Wells (1986: 416) also argue that non-glottalised released /t/s are associated with the end of a conversational turn in Tyneside English. The phenomenon of the almost categorical realisation of (t) as a fully released stop with no glottalisation in pre-pausal and turn-final positions has been termed the Pre Pausal Constraint (PPC) (Milroy 1997) or the Final Release Rule (FRR) (Docherty et al. 1997). Indeed, in the data from Newcastle, the FRR was found to apply 99.8% of the time in word list readings. In conversational data full release was found to occur in the majority of cases, but violations of the rule were observed, which will be considered in more detail presently.

Little evidence exists of the geographical or sociolinguistic distribution of the unreleased stop, [f], It is noted as a feature of RP, however, specifically in turn- 103 final position (see O’Connor 1973: 134). Although ‘[t]he non-release of final plosives is a feature of colloquial RP’ (Cruttenden 1994: 145), use of [f] in RP may be subject to stylistic variation, as Cruttenden (1994: 145) adds that ‘[c]areful speakers...tend to release such plosives audibly, and those who, in ordinary conversational style, use the unexploded variety will often use an audible release in more formal circumstances.’

The glottal stop realisation of (t) in a pre-pausal environment is widespread in BrE and has recently been included as a feature of RP (Wells 1990, 1994, 1997). However, Wells states that this is a feature found ‘in casual RP’ (1990: 6) and ‘[ajmong younger RP speakers’ (1994: 201). (t)-glottalling in pre-pausal position, then, is, according to Wells, a more recent change than (t)-glottalling before obstruents and sonorants which began in RP in the mid-twentieth century (1997: 21). Fabricius (2000), who carried out a detailed study of the distribution and acceptability of (t)-glottalling in RP, found that (t)-glottalling in a pre-pausal environment is in evidence in interview style but is used to a considerably lesser degree than in pre-consonantal environments (36% use recorded overall). Moreover, (t)-glottalling was found to be subject to dramatic style shifting, with pre-pausal glottal replacement virtually absent from reading passage style.

(t)-glottalling in pre-pausal position is claimed as a feature of London English

(Hughes & Trudgill 1996: 70). Indeed, [?] is found to be ‘near categorical' pre- pausally in South East London English, but found ‘only sporadically’ amongst older South East London Regional Standard speakers (Tollfree 1999: 171). Altendorf (1999), in her study of three socially stratified London schools, found high rates of (t)-glottalling in pre-pausal environments across all three groups in interview style. However, pre-pausal (t)-glottalling was subject to style shifting amongst the two groups of public school informants, whilst the comprehensive

school informants continued to use [?] in the more formal reading style.

Elsewhere (t)-glottalling in pre-pausal environments is reported as the ‘majority pronunciation’ in Derby, particularly amongst young speakers, with an 81 % use noted for younger speakers (Docherty & Foulkes 1999: 50). In the West Midland 104

borough of Sandwell, Mathisen (1999: 116) finds a 16% use of pre-pausal [?],

claiming ‘[gjlottaling before a pause or vowel is usually considered nonstandard’. In contrast, in Cardiff English glottalisation is considered a prestige form, with glottal replacement of pre-pausal (t) appearing more in MC female speech than WC female speech (Mees & Collins 1999: 196). In Glasgow,

Stuart-Smith (1999: 196) finds [?] to be ‘obligatory’ for (t) in WC speech across age groups. Glottal stop realisations of pre-pausal (t) are also found occasionally in Newcastle English, as noted, and these will be considered in more detail within the course of this section of the chapter.

As with [ht] and [tn], little evidence exists for the distribution, both geographical

and social, of 0 for pre-pausal (t). One study which includes the zero realisation of (t) amongst the variants under investigation is that of Cardiff English (Mees and Collins 1999). Interestingly, 0 is found to be the preferred variant of WC females (67.6% use recorded from 1981 data). (Recall that the glottal stop is considered by Mees and Collins (1999) to be a prestige form in Cardiff English and a higher level of usage was recorded in MC female speech.) However, the level of use of 0 was found to have decreased dramatically by 1990 (28.6% use) in the speech of four subjects.

4.1.3 Particulars of the data and areas of interest

In the conversational data, 30 tokens of both pre-pausal (t) and turn-final (t) for all speakers were sought. The chosen environment was post-vocalic, for example, that, it, but, as it was felt that a post-vocalic environment would be the clearest in which to undertake an auditory analysis of the tokens under

consideration, particularly the identification of [ht] (see Fabricius 2000: 81 for

similar justification). In citation form, 14 tokens of pre-pausal (t) were elicited per speaker. These included single word citation forms (for example, gate, goat, sheet) and sequences ending in word-final (t) (for example, I bought it, I beat it). Due to a lack of a break in connected speech, many of the word-final tokens of (t) could not be classified as pre-pausal. As tokens classified as pre-pausal only 105 were analysed, for the majority of informants, fewer than the 14 tokens in citation form were included in the results.

4.1.3.1 Some discourse considerations

The findings from the Middlesbrough data will be compared closely with the findings from recent studies of Tyneside English. As the FRR can be seen as being operational in both Tyneside English and Durham English, whether constraints on variation due to conversational structure are in evidence in the data from Middlesbrough is of primary interest. The decision to separate the data discoursally into pre-pausal environments and turn-final environments was motivated by the desire to ascertain whether discoursal constraints were operating in use of [?] realisations in the data. This separation is not unproblematic, however.

A pre-pausal (t) is identified as one occurring immediately prior to a break in connected speech of usually one second or more. However, if a break in connected speech is followed by another speaker taking the floor, whether the pause can be viewed as turn-final or pre-pausal is not always indubitable. In conversational structure, the basic unit of analysis is customarily recognised as the turn. A turn-final (t) in the present study, then, is taken as a token of (t) occurring at what ethnomethodologists and conversational analysts term a Transition-Relevant Place (TRP) in the discourse (Sacks, Schegloff and Jefferson 1974). A TRP occurs at a recognisable end of a turn-construction unit, a point in the discourse at which a speaker has finished his/her turn and is ready to yield the floor to another speaker. Providing the turn could stand alone as a meaningful utterance, the TRP could appear after a turn of as little as one word.

However, if a turn is taken to indicate everything a speaker says before another speaker begins to speak, that is, the ‘one speaker at a time model’ (Sacks, Schegloff and Jefferson 1974, Schegloff 2000), instances of simultaneous speech in the discourse pose immediate problems. This simultaneous speech can take the form of interruptions, which are frequently classified as disruptive (see 106

Zimmerman and West 1975). Interruptions, or ‘disruptive turns’ (Holmes 1995: 52), are competitive discourse strategies which signal a battle for management of the floor. If the interruption is successful, that is, if the current speaker is prevented from finishing his/her turn, the problem arises of whether the token is classified as pre-pausal (given that the speaker did not intend to yield the floor) or turn-final (as in the ‘one speaker at a time’ model). Additional problems are posed by instances of simultaneous talk which can be identified as occurrences of two speakers completing an utterance simultaneously but differently, or two speakers saying the same thing at slightly different times (see Coates 1996: 128). Similarly, overlaps represent instances of simultaneous speech. Coates (1993: 109) defines overlaps as ‘instances of slight over-anticipation by the next speaker’. These overlaps are generally uttered at TRPs, further confounding the problem of determining when a turn can be said to be finished and thus when a (t) can be said to be turn-final.

TRPs are usually signalled in advance, possibly semantically, but more commonly through intonational and paralinguistic features. Semantic evidence and intonational features were used to assess whether a token appeared at a TRP. Tokens occurring at points wherein the speaker was felt to be using strategies to keep the floor, for example, avoiding pausing at a natural break, rather pausing at a point at which the message was incomplete, were thus classified as pre-pausal. Topic negotiation and shifts were also used to indicate ends of turns. Additionally, use of discourse markers was taken into account although not used in a categorical manner. Use of question tags, isn’t it?, doesn 't it?, haven ’t you?, and sentence tags, and that, stuff like that, are often taken to indicate the end of a turn. Indeed, tags were found to be the main intra-linguistic factor affecting the

use of turn-final [?] in the Newcastle study, as violations of the FRR were

‘accounted for with reference to the turn-delimitative functions of the FRR, being redundant when it co-occurs with a sentence tag which is a particularly salient turn-delimitative cue’ (Docherty et al. 1997: 299). Thus, in Tyneside English, the turn-delimitative function of tags is seen as something of a Trojan Horse ‘sneaking the innovatory glottal realisation into the last relevant context of the system which has hitherto resisted it’ (Milroy 1997: 9). However, tags do not 107 automatically signal the end of a turn: numerous examples of tags appearing in pre-pausal position, and also where there is no break in connected speech, are found in the data. The minimal responses right, that’s right which can signal active listenership, or, if delayed, can signal a lack of interest or agreement, similarly posed problems in terms of whether they constituted a turn and therefore the end of the previous speaker’s turn. Tags and minimal responses in pre-pausal and turn-final position were analysed, and findings revealed marked sociolinguistic patterning which will be discussed briefly in section 4.1.6.

No one single criterion was used, then, when defining a token as pre-pausal or turn-final. A decision was made on each instance with reference to intonational, semantic, topic-based and discourse marker clues. Mindfulness of the speaker’s intention to keep or yield the floor was central to the classification. However, as precise knowledge of the speaker’s intention is not accessible to the analyst, the decision is always necessarily somewhat subjective.

We turn now to the findings. Before correlating the data with particular speaker groups, the overall distribution of the variants of (t) in pre-pausal and turn-final positions in conversational style are presented. Following this we turn to an analysis of the data broken down by age, then gender, then age and gender. Examination of the distribution of variants of (t) in pre-pausal position in word- list style will follow, and finally the sociolinguistic patterning found in use of the discourse strategies under consideration will be presented.

4.1.4 Overall distribution of variants of (t): conversational style

Table 4.1 below presents the overall use of the five variants of (t) in pre-pausal position from the conversational data from Middlesbrough. 108

Table 4.1: Overall use of variants of (t) in pre-pausal position (conversational style, all speakers)

Iht] It] m [?] 0 Total n. % n. % n. % n. % n. % 911 36 4 280 30.7 30 3.3 537 58.9 28 3.1

Over half of the tokens of pre-pausal (t) taken from all speakers in conversational style are accounted for by glottal stop realisations. This result is markedly different from the ‘occasional violations of the FRR’ found in the Newcastle study (Docherty at al. 1997: 296). [t] is the variant with the second highest usage in the overall scores. However, at 30.7%, use of [t] is considerably lower than use of [?] in the data, unlike the results found in Tyneside. Overall, [?] and [t]

realisations make up the vast majority of the tokens analysed, with their combined scores accounting for 89.6% of the data. The remaining 10.4% of the data are distributed virtually equally amongst the other variants. Use of [ht], [f]

and 0 is marginal in comparison with use of [?] and [t], with a slightly higher

score noted in the level of usage of [ht].

Whether use of the variants of pre-pausal (t) correlates with particular speaker groups will be revealed in the course of the section. We turn first to an analysis of the overall distribution of variants of turn-final (t) in conversational style as presented in Table 4.2 below.

Table 4.2: Overall distribution of variants of (t) in turn-final position (conversational style, all speakers)

[ht] [t] It’] 1?] 0 Total n. % n. % n. % n. % n. % 932 37 4 397 42.6 10 1.1 460 49.3 28 3

fable 4.2 reveals that the overall preferred variant of turn-final (t) in the data is also |?]. Additionally, as found for pre-pausal (t), [t] is the variant exhibiting the

second highest score. Also, as was the case with pre-pausal (t), [?] and (t| usage 109 accounts for the vast majority of the data (91.9%). Differences are also revealed between the data from Middlesbrough and the data from Newcastle, where glottal realisations in turn-final positions were found to be rare. Although use of [?] is lower in turn-final position than in pre-pausal position in the

Middlesbrough data (49.3% and 58.9% respectively), the extent of the use of [?] in turn-final position contrasts markedly with the use found in the Newcastle data and suggests that the FRR is not operating in MbE. Nonetheless, as differences between use of [?] and [t] in pre-pausal (t) and turn-final (t) are in evidence in the

Middlesbrough data, it appears that the choice of variant is affected by the discoursal position of the token. Marginal use of the other three variants also is demonstrated in the data for turn-final (t), with a slightly higher level of use of |ht] again revealed. We turn now to an analysis of the data broken down into the particular speaker groups, beginning with (t) in pre-pausal position.

4.1.5 Pre-pausal (t), conversational style

4.1.5.1 Generational variation

Figure 4.1 below shows the distribution of variants of pre-pausal (t) across age groups. The conflated scores of the young adults and the adolescents arc presented in the young group.

Figure 4.1: Pre-pausal (t), generational differences

80 78.3

70

60 58.2

■ pre-aspirated ■ released eP P unreleased □ glottaled ■ zero

0 old (N. 220) middle (N. 216) young (N 475) 110

Considerable variation in the distribution of variants of pre-pausal (t) across age is revealed. The data suggest that [ht] is a new variant in MbE, it being in evidence in the speech of the young of the sample but not in the old. Although some use of [ht] is shown in the speech of the middle group, this amounts to three tokens from 216 (all three tokens were produced by one female speaker from the middle group). Use of [t] is revealed to be in dramatic decline in the data. Use of |f | is also revealed to be decreasing across age, with use only really in evidence among the old speakers. The middle group demonstrates some use of [t’J, but this amounts to six tokens (one each from three female speakers and one each from three males). Apart from [ht], then, the only other variant to demonstrate an increased use across age in the data is [?], which rises steadily, reaching its highest point of 78.3% in the speech of the young of the sample. Finally, apart from [t] and [?], the zero realisation is the only other variant of which some use, albeit marginal, is recorded across the three age groups. A decline in use of 0 is exhibited across age in the data, however. 0 reaches its lowest level of usage in the speech of the young, where use amounts to five tokens from 235.

4.1.5.2 Gender

Figure 4.2 : Pre-pausal (t), gender

70 63.3 60 54 8

50

■ pre-aspirated 40 a released □ unreleased SB 30.5 □ glottal stop 30 ■ zero

20

10

male (N 449) Ill

Gender-correlated variation in the distribution of variants of pre-pausal (t) is in evidence, with female speakers appearing to make more extensive use of the variants available to them. Additionally, female speakers are revealed to be responsible for the use of [ht] in the data, with only two tokens of [ht] recorded

from a total of 449 tokens from male speakers. Interestingly, use of [t] amongst males and females is equal. A higher level of use of [f] is exhibited by the males

than the females. However, use is low across the sample. Although males also demonstrate a higher use of [?] than the female speakers, the difference is not

statistically significant, and [?] is revealed to be the preferred variant of pre-

pausal (t) for both male and female speakers. A zero realisation is found to be used more frequently by female speakers than males of the sample, with only six tokens recorded from males. Also it is interesting to note that all five variants are used to some degree by both groups of speakers. Whether this will be found to be true when the data are broken down by age and gender is revealed below.

4.1.5.3 Age and gender

f igure 4.3 presents the distribution of variants of pre-pausal (t) for the male speakers broken down by age.

Figure 4.3: Pre-pausal (t), male speakers

old male middle young adolescent (N. 105) male(N. adult male male (N. 109) (N. 120) 115) 112

When the data are broken down by age and gender, we see that certain variants arc avoided categorically by speaker groups. Indeed, in no male speaker group is use of all five variants demonstrated. The old male speakers reveal the most variability in their use of variants of pre-pausal (t). The young adult males show the least variability, with only two variants effectively used in the speech analysed (the 0.8% use of [ht] recorded for the young adult males is accounted for by one token from one member of the group).

In use of the two variants which make up the majority of the data, [t] and [?], we see a steady decline in the use of [t] and a steady increase in use of [?] across age f rom old to young. Use of [t] decreases considerably from the preferred variant of all four members of the old group to categorical rejection by two of the adolescent males. Use of [?] mirrors use of [t] in that a dramatic increase across age is recorded in the data. A near categorical use is recorded in the speech of the adolescent males. This is compared with a 5% use of glottal or globalised tokens of pre-pausal (t) recorded for the young WC males in the Newcastle study (Docherty et al. 1997: 296). In stark contrast, in the Middlesbrough data [?] is the preferred variant of the middle, young adult and adolescent male groups. Furthermore, [?] is the preferred variant of all eight members of the two young male groups. It is, in fact, used categorically by three of the young males (one young adult and two adolescents).

Use of [ht] is found in the speech of the young males of the sample only. Use of

|'Ml in the male data, however, amounts to two tokens from 235, (one recorded from a young adult male and one from an adolescent male). Use of [f] is largely confined to the speech of the old males of the sample, with all members of the group making some use of this variant. Use declines considerably between the old group and the middle group, in which only three tokens are recorded, [f ] is then categorically avoided by all eight young males. The zero realisation of pre- pausal (t) is found to a very limited extent in the speech of the males. Interestingly, the adolescent group shows the highest use of 0 . However, this amounts to three tokens only. 113

Whether the trends uncovered in the male data are duplicated in the speech oi the females is revealed below. Figure 4.4 below presents the distribution ol variants of pre-pausal (t) amongst the female speakers.

Figure 4.4: Pre-pausal (t), female speakers

80 77.5

70 65

60 53.9 ■ pre-aspirated 50 45.8 ■ released 42.1 □ unreleased ^ 40 □ glottaled 32.2 nzero 30 21.7 20

10 7.8 3.3 1.6

old female middle young Iadolescent (N. 115) female (N. adult female (N 107) female (N. 120) 120)

Female speakers are revealed to make more use of the full set ol variants under consideration than do the males. Nevertheless, it is only in the speech of the middle group of females that use of all five variants is revealed. However, use of four of the five variants examined can be found to some extent in the speech of the old, young adult and adolescent female groups.

The same trends as those found in the male data are demonstrated in use of the two variants which account for the bulk of the data, [t] and [?]. [t| secs a sharp decline and [?] exhibits a considerable increase across age from old to young.

Use of [t] does not decrease as dramatically as in the data from the male speakers of the sample. Although [t] is the preferred variant of the old female speakers, it is the preferred variant of two of the old females only. A higher use of [t] is demonstrated in the speech of the old males where it is the preferred variant of all four members of the group. Additionally, the two old female speakers who demonstrate a preference for [t] are the two oldest members of the group (ages 69 114 and 80). Both o f the 60 year old females reveal [?] to be their preferred variant.

Unlike the middle males, speakers from the middle female group also reveal |t) to be their preferred variant. Use o f [t] then declines considerably, reaching its lowest level of use amongst the adolescent females.

Although the same trend is revealed in use o f [?] in the female data compared with the male data, the increase in [?J is not as dramatic. The old female speakers show a substantially higher use o f [?] than their male counterparts (32.2% compared with 19%). However, 83.8% (31 o f the 37 tokens) o f [?] recorded in the speech of the old females was accounted for by the two 60 year old informants. Use o f [?] reaches its highest point in the speech o f the young adults in the female data. Use is lower than that found in the speech o f their male counterparts, however. Interestingly, use of [?] then declines considerably in the speech o f the adolescent females, who show a use o f [?] for pre-pausal (t) which is significantly lower than the male adolescents' use (p<0.049). However, use of

| VI amongst the young females of the sample is still markedly higher than that found in the speech of the young WC females in the Newcastle study, who demonstrated a considerably higher use o f pre-pausal [?] than any other group in the Newcastle sample. (A 30% use was recorded from the young WC females in Newcastle.)

I he considerably lower use o f [?] recorded from the female adolescents compared with the other three young speaker groups is due. in part, to the adolescent females’ higher use o f the innovatory variant in the data. (ht).

Although [ht] is in evidence in the speech o f the middle and young adult females, only three tokens are recorded from the middle group (from the same speaker) and live tokens from the young adult speakers (three from one speaker and one each from two other speakers o f the group). In contrast, all four o f the adolescent females use |ht] to some extent in pre-pausal position, with one speaker revealing

I llt| to be her preferred variant. The four female adolescents account for 72.2%

(26 of the 36 tokens recorded) of the [ht] realisations from the whole sample of 115 speakers. Thus, the same sociolinguistic patterning in use of [ht] is revealed as that found in Newcastle. (Recall that a much higher use of [ht] was recorded from the young female speakers in Newcastle than the males, although the young age group in the Newcastle study ranged from 16-25 and no indication of whether any difference is found between female adolescents and female young adults is given in the findings.)

[tn] is in evidence in the speech of the females, although use is lower than that recorded in the male data. The old female speakers reveal the highest level of use of [f], with all four members of the group using the variant to some extent. Use of [f] declines in the middle group and is then rejected categorically by the young females, as was found to be the case with the young males. Also a higher use of 0 is revealed in the female data compared with the male data. Use of 0 declines steadily from old to young, reaching its lowest point in the speech of the adolescents, where only two tokens are recorded.

In order to ascertain whether position in the discourse affects the choice of variant used, we turn to a comparative analysis of the distribution of variants of (t) in turn-final position.

4.1.6 Turn-final (t), conversational style

4.1.6.1 Generational variation

Figure 4.5 below presents the distribution of variants of turn-final (t) across age with the young adult and adolescent scores combined into one group. 116

Figure 4.5: Turn-final (t), generational differences

80

old (N. 232) middle (N. 232) young (N. 468)

The same general trends are revealed in Figure 4.5 in the generational distribution of variants of turn-final (t) as those found in the distribution of variants of pre-pausal (t) as presented in Figure 4.1. The majority of the data is accounted for by two variants, [t] and [?], one of which, [t], appears to be declining dramatically whilst the other, [?], appears to be increasing dramatically across age from old to young.

Although the same declining trend is revealed in turn-final (t) as in pre-pausal (t), a higher level of use of [t] in turn-final position than in pre-pausal position is in evidence in all three age groups, suggesting that the discoursal position of the variable does affect the choice of variant used. Likewise, use of [?] mirrors use of [t] across age as was found in pre-pausal (t). Nonetheless, a higher level of use o f [?] for (t) in pre-pausal position than in turn-final position is revealed in all three age groups. Thus the data suggest that [t] is more likely to be used if the (t) is turn-final whilst [?] is more likely to be used if (t) appears in pre-pausal position.

As with the data for pre-pausal (t), we find marginal use of the other three variants, [ht], ft'1] and 0 , compared with use of [t] and [?]. The data suggest, again, that [ht] is new to MbE, appearing as it does almost exclusively in the 117 speech of the young. (A 0.9% use of [ht] is recorded from the middle age group, which is accounted for by two tokens.) Furthermore, use of [ht] in the speech of the young is slightly higher in turn-final position that in pre-pausal position. Use of [tn] seems also to be confined to the speech of the old group, with only one token recorded from the middle group. Use of [f] is higher in turn-final position than in pre-pausal position. Use of 0 declines across age as it did for pre-pausal

(t)-

4.1.6.2 Gender

Figure 4.6 presents the distribution of variants of turn-final (t) from the conversational data broken down by gender.

Figure 4.6: Turn-final (t), gender

60 54.1

50 44.7

40 mu pre-aspirated ■ released 30 □ unreleased □ glottaled H zero 20

10 5.3

male (N. 458) female (N. 474)

The female speakers appear to make more use of the variants available to them than do the male speakers, as was found in the data for pre-pausal (t) in Figure 4.2. Again, it is the female speakers who are found to be virtually wholly responsible for tokens of [ht] in the data, with only a 0.2% use of [ht] noted in the male data (one token from 458). As in the data for pre-pausal (t), use of [t] is almost exactly the same for male speakers as it is for female speakers, which is interesting given the gender-correlated variation found in use of all other variants. Use of [t] is approximately 12% higher in turn-final position than in 118 pre-pausal position for both males and females. Use of [f] is lower than in pre- pausal position and is confined to the male speakers in turn-final position (unlike in pre-pausal position where the females revealed some use). [?] is the preferred variant for both males and females again. The males reveal a higher use, however. Although the trends are the same, use of [?] is about 10% lower for both males and females in a turn-final environment than in a pre-pausal environment. 0 is more frequent in the female data than in the male data, which was the pattern found in the pre-pausal data. Use is low for both groups, although interestingly it is slightly higher in turn-final position than in pre-pausal position. Whether the same trends are revealed when the data are broken down into the particular speaker groups selected is now considered.

4.1.6.3 Age and gender

Figure 4.7 presents the distribution of variants of turn-final (t) from the male data broken down into the four age groups of old, middle, young adult and adolescent.

Figure 4.7: Turn-final (t), male speakers

80 77 77.5

70

60

1 pre-aspirated 50 I released □ unreleased 3? 40 □ glottaled 30 □ zero 21.7 20

10 oj< 0 old male middle young adolescent (N. 112) male (N. adult male male (N. 113) (N. 113) 120)

The familiar decrease in [t] and increase in the use of [?] across age from old to young is revealed in Figure 4.7. However, there is less variability in the data for turn-final (t) than in that found for pre-pausal (t) (see Figure 4.3), with an 119 extremely low use exhibited of the other variants under consideration. The old males, again, demonstrate the most variability in their speech.

As regards [t], a sharp and steady decline in use from old to young is revealed. Its lowest point of 21.7% is considerably higher than the 9.6% reached in the pre- pausal data, however. Likewise, use of [?] reveals the same trend as that found in the pre-pausal data. Use is lower in turn-final position than in pre-pausal position for all speaker groups. [?] is still the preferred variant for all members of the adolescent group, however, and is compared with an 11 % use by young WC males in the Newcastle study. (This figure represents one token of nine (Docherty et al. 1997: 297).)

[f] is the only other variant of which a use of over 2% is demonstrated in turn- final position. The old male speakers exhibit an 8% use of [f], with all four members of the group showing some use of the unreleased variant. Only one other token of [tn] is recorded in the data from a middle male speaker. Use of [f ] in turn-final position found in the old group is half that demonstrated by the same group in the pre-pausal data. One token only of [ht] is recorded in turn-final position from an adolescent speaker, and a very limited use of the zero realisation is also shown in the data, with one token noted in the old group and two recorded from the middle group. 0 is categorically avoided by the male speakers.

Figure 4.8 below reveals whether the same trends are found in the speech of the females of the sample.

The female speakers are revealed to make more use of the variants available than the male speakers, yet the variability in the speech of the females is not as widespread as that found in the female pre-pausal data as presented in Figure 4.4. None of the groups of female speakers uses all five variants of turn-final (t). 120

Figure 4.8: Turn-final (t), female speakers

70 70

DU pre-aspirated ■ released □ unreleased □ glottaled zero

0 old female middle young adolescent (N. 120) female (N. adult female (N. 119) female (N. 115) 120)

Again, the majority of the data is accounted for by [t] and [?]. The pattern found in pre-pausal (t) as presented in Figure 4.4 is duplicated in the findings for turn- final (t) in the female data, with use of [?] declining in the adolescent group after a steady increase from old to young adult. Although the difference between young adult females and adolescent females is not as great in turn-final position as in pre-pausal, a decline in use is still noted. Furthermore, use of [?] is still markedly higher than that found in the speech of the young WC females in the Newcastle study (39%), who again demonstrated a considerably higher use of [?] than any other speaker group in the Newcastle study. The old female group reveal a higher use of [?] than the old males, which was also the pattern found in the pre-pausal data. Again, an age difference appears to be discernible within the old female group of speakers, with one of the younger members of the old females (aged 60) accounting for 64.3% of the [?] score (18 of the 28 tokens recorded for the old female group). In all other age groups, use of [?] is higher in the male data than in the female data, which again is precisely the pattern found in pre-pausal (t). Additionally, use of [?] in turn-final position is lower than that in pre-pausal position for all age groups of females.

A steady decline is noted in the use of [t] from old to young. The old females and the adolescent females demonstrate a lower use of [t] than that of their male 121 counterparts. The middle group and the young adult group of females exhibit a higher use of [t] than the middle and young adult males, however. Also, the old, middle and young adult female groups demonstrate a considerably higher use of [t] in turn-final position than in pre-pausal position. This is not found to be true of the adolescent female group.

The fact that the adolescent females demonstrate a declining use of [?], yet also a much lower use of [t] compared with the other female groups, is explained by the adolescent female use of the innovatory variant, [ht]. The adolescent female speakers demonstrate a 25.2% use of [ht], with all speakers of the group using the variant to some extent. Furthermore, three of the adolescents are found to have a higher level of usage of [ht] than [t]. The 25.2% adolescent female use of [ht] is compared with a 4.2% use (five tokens) in the speech of the young adult females, a 1.7% use (two tokens) in the speech of the middle females and a 0.8% use (one token) in the speech of the adolescent males. The adolescent females, then, are the speakers of the sample who are overwhelmingly responsible for the introduction of the innovatory [ht] into the conversational data, as was found to be the case for pre-pausal (t).

[tn] is categorically rejected in turn-final position by all groups of female speakers, despite marginal use by the old female group and the middle female group in pre-pausal position. In contrast, 0 is found in all four age groups of females. Use of 0 declines across age in the female data, however, reaching its lowest point in the adolescent speech, from which only three tokens are recorded. Although use of 0 is low in the female data, it is still higher than that found in the male data.

We turn now to an examination of the distribution of variants of pre-pausal (t) in word-list readings. 122

4.1.7 Pre-pausal (t): citation forms

Fourteen tokens of pre-pausal (t) were included in citation form to ascertain stylistic variation in the variants under consideration. Due to the lack of a discernible break in connected speech, not all tokens were included in the analysis, as noted in section 4.1.3. Table 4.3 below presents the overall distribution of variants of pre-pausal (t) in the formal reading style.

Table 4.3: Overall distribution of variants of pre-pausal (t): citation forms

[ht] ft] [t] m 0 Total n. % n. % n. % n. % n. % 289 6 2.1 220 76.2 1 0.3 61 21.1 1 0.3

Over three quarters of the tokens in word-list style are realised as [t]. This stands in stark contrast to the overall figure of 30.7% in conversational style (see Table 4.1). Use of [?] in reading style is considerably lower than in conversational style

(21.1% and 58.9% respectively). This would suggest that use of [?] is subject to

style shifting. The three other variants under consideration are avoided almost

categorically in word list style (one token recorded each of [f ] and 0 , and six

tokens of [ht]). Whether any sociolinguistic patterning is in evidence when the

data are broken down into the particular speaker groups is considered below. 123

Table 4.4: Distribution of variants of pre-pausal (t) by age and gender: word list style (WLS) and conversational style (CS)3

[ht] % [t] % [t 1 % [?] % 0 %

WLSCSWLSCS WLSCS WLSCS WLS CS

Om 0 0 96.8 62.9 3.2 16.2 0 19 0 1.9

Mm 0 0 86.5 38.5 0 2.8 13.5 57.8 0 0.9

YAm 0 0.8 51.4 15 0 0 48.6 84.2 0 0

Adm 0 0.9 69 9.6 0 0 31 86.9 0 2.6

O f 0 0 100 53.9 0 6.1 0 32.2 0 7.8

M f 0 2.8 100 45.8 0 2.8 0 42.1 0 6.5

Y A f 0 4.2 51.4 15 0 0 45.9 77.5 2.7 3.3

A df 23.1 21.7 50 11.7 0 0 26.9 65 0 1.6

As Table 4.4 reveals, all groups of speakers show a considerably lower use of pre-pausal [?] in word-list style than that found in conversational style. In most

cases use of [?] is less than half that found in conversational style. Indeed, the old

males, old females and middle females categorically avoid [?] in word-list style.

The highest level of use of [?] is found in the speech of the young adult males.

Even in this group, however, use of [?] is considerably lower than in conversational style. The young adult females show a comparable level of use of

3 The abbreviations used correspond to the following: Om (Old male), Mm (Middle male), YAm (Young Adult male), Adm (Adolescent male), Of (Old female), Mf (Middle female), YAf (Young Adult female) and Adf (Adolescent female). 124

[?]. Use then declines in the speech of the adolescents, however. This was the pattern found in the conversational data for the females. In the older speaker groups, [?] realisations are only in evidence in the speech of the middle males.

However, the figure of 13.5% is accounted for by seven tokens of [?], all of which were produced by the same speaker, [t] is the preferred variant of all speaker groups, with the lowest level of use found in the speech of the adolescent females. The old and middle female groups reveal categorical use of [t] in word- list style despite showing 53.9% and 45.8% use of [t] in conversational style.

A 23.1% use of [ht] is revealed in the speech of the adolescent females, which is slightly higher than the level recorded in the conversational data. It must be noted, however, that tokens of [ht] in word-list style were largely accounted for by one of the adolescent females (five of the six tokens). Marginal use of the other variants is revealed in the data. [tn] and 0 scores in the data are accounted for by one token each. The same patterns, then, are revealed in the word-list data as those found in the conversational data for pre-pausal (t). It appears that use of [?] in this environment is subject to style shifting, suggesting that pre-pausal [?] is above the level of awareness and is a marker of variation to the speakers of the sample.

4.1.8 More discourse considerations

As noted in section 4 .1.3.1, the data were analysed for use of discourse markers (minimal responses, question tags and sentence tags) in pre-pausal and turn-final position. Whilst it is not within the scope of this work to analyse both the form and the function of the features under consideration, interesting findings were revealed, and we turn now to an examination of the sociolinguistic patterning found in the data when analysed for discoursal patterns.

In terms of the use of minimal responses, a considerably higher incidence of this discourse strategy was demonstrated by the male speakers of the sample than by 125 the female speakers. This is interesting in itself, as female speakers have been found to make more use of minimal responses than males in many previous studies. Indeed, Coates (1986: 102) claims ‘[rjesearch on the use of minimal responses is unanimous in showing that women use them more’.4 Nonetheless, in the speech of the two young age groups seven minimal responses are recorded from male speakers as compared with one from a female speaker (seven of the eight minimal responses found in the speech of the young are recorded with [t] realisations). Furthermore, in the speech of the older groups, one group of speakers is conspicuous for the much higher number of minimal responses recorded in the conversational data than any other group, that group being the middle male group. 29 instances of minimal responses are recorded from the speech of the middle males as compared with five from the middle females, one from the old males and eight from the old females. The majority of the minimal responses (33 of 43) recorded from the middle and old speaker groups have [t] realisations of (t), nine [?] are noted and one [ht].

Likewise, an analysis of the use of tags reveal interesting sociolinguistic patterning. Again, use of tags has been associated with female speech. Lakoff (1975: 16) suggests that tags are considerably more likely to be used by women than men, as women are generally more tentative in their assertions. Although studies have confirmed that speakers assume a connection between tags and female speech, few studies have proved this (see Coates 1986: 104). However, young WC females were found to use by far the highest number of tags in the Tyneside data, which was posited as an explanation for the higher female level of violation o f the FRR (Docherty et al. 1997: 299). The same pattern cannot be said to exist in the Middlesbrough data, however. Overall, the male speakers demonstrate a substantially higher use of tags than their female counterparts5.

4 It must be remembered that, as the interest of the present study is in the realisation of pre-pausal and turn-final (t), the only minimal responses included in the analysis are ‘right’ and ‘that’s right’. Again, it should be borne in mind that tags ending in a word-final /t/ only are included in the analysis. 126

Table 4.5 below presents the level of use of question tags and sentence tags6 found in particular speaker groups.

Table 4.5: Number of question tags and sentence tags recorded by age and gender

Question tags Sentence tags Turn-final Pre-pausal Turn-final Pre-pausal Om 11 6 1 0 Mm 14 3 0 0 YAm 14 5 13 20 Adm 15 10 10 4

O f 10 4 1 3 M f 18 8 0 0 YAf 14 1 2 0 Adf 16 4 1 0

Question tags are revealed to be used to a considerable extent by all speaker groups, whilst sentence tags have a much more limited use. Question tags appear much more likely to occur in turn-final position (particularly in the female data) which substantiates, to a degree, their function as a turn-delimitative cue. However, a considerable number are recorded in pre-pausal position also. Overall, the male speakers reveal a slightly higher level of use of question tags than do females (78 and 75 respectively). Question tags appear fairly evenly spread across age and gender, with the middle female group and the adolescent male group revealing the highest levels of use.7

6 Sentence tags included and that, and all that, stuff like that, something like that. 7 Innit? as an invariant tag is seen as a fairly recent innovation of London English (see Anderson 1997). Although nine instances were noted in the pilot study of Middlesbrough (Llamas 1998), only one instance was recorded in the speech of a young adult male in the present study due, perhaps, to the difference of formality between an interview context and a free conversation context. 127

Sentence tags are surprisingly uncommon in the data. Nonetheless, one group of speakers exhibits a much higher level of use than any other speaker group. 23 sentence tags are noted in the data from the young adult males. Furthermore, 14 are recorded in the sample from the adolescent males. No other group of speakers makes an equivalently high use of this discourse strategy.

The evidence presented also bears out the association of use of [?] and tags amongst younger speakers. 11 of the 16 young speakers demonstrate categorical use o f [?] for final (t) in tags ([t] or [ht] forms are recorded in 12 o f 129 tokens of

(t) (9.3%) in tags in the data)8. This is compared with a 40.5% use (32 of 79) of [?] in tags recorded in the middle and old groups. Use of both question tags and sentence tags has increased considerably in the speech of the two young groups compared with use in the two older groups (129 in younger and 79 in older). This is combined with an increase in [?] realisations of pre-pausal and turn-final (t) in tags. It is not within the scope of this work to analyse differences in function between the discourse markers analysed, for example, gender differences in use, connection between use of tags and the speaker’s intention to keep or yield the floor. However, a more thorough conversational analysis of the sociolinguistic patterning of discourse strategies is planned as a future work.

4.1,9 Pre-pausal and turn final (t): summary

A picture of complex variation and sociolinguistic patterning has been revealed in the data for pre-pausal and turn-final (t). Gender-correlated variation is in evidence as is variation across age groups, which is suggestive of change in progress. Although the same sociolinguistic patterning is revealed in both discoursal environments, differences between the distribution of variants of pre- pausal and turn final (t) are discernible, indicating that choice of variant is affected by discoursal position, [t] is revealed to be more likely to occur if the

s Interestingly, the use of lengthened Ini was found in question tags amongst the young male speakers. Five males (two adolescents and three young adults) produced 18 instances of lengthened /n/s on the lexical items innit?, dunnit?, hannit? dinnit? 13 of the 18 were produced by one speaker (adolescent). No instances were noted in the speech of the females. 128 token appears in turn-final environment, whilst [?] is the more likely realisation of (t) when it occurs pre-pausally. Additionally, in terms of discoursal features, strategies such as use of minimal responses and tags have been found to be more prevalent in male speech than female speech, which runs contrary to conventional wisdom.

Across age in both discoursal environments, a marked decline in use of [t] and increase in use of [?] is apparent. Although a marked preference for use o f [?] is revealed over use of the disfavoured [t] in the speech of the young, near categorical use of [t] by the old speakers is not shown in either discoursal position, unlike in Tyneside English. Therefore, the FRR cannot be seen as being operational in MbE as it is in Tyneside and Durham English. Furthermore, no evidence exists of an increase in use of the localised [t] in this environment, an increase which would be suggestive of the convergence of MbE with varieties further north. However, an interesting decrease in the use of [?] amongst the adolescent females as compared with the young adult females in both discoursal environments is noted. Despite [?] being subject to style shifting, it may be that the stigma attached to the glottal replacement of (t) is lessening as use becomes increasingly widespread, with the result that ‘[wjhat started as a vulgarism is becoming respectable’ Wells (1994: 201). This may entail the avoidance of the glottal stop realisation of (t) amongst young speakers who value the covert prestige of non-standard forms. The evident disfavouring of [?], then, will perhaps prove a fruitful line of future enquiry.

Although every member of every group of speakers shows some use of [?], the males in the sample demonstrate a higher level of use than the females. This is contrary to the findings of the Tyneside study, where a higher level of use of [?] was found in the speech of the young females, who were the speakers found to be responsible for the majority of FRR violations. A higher male use of [?] is in keeping with the association of [?] as a WC male norm. However, a considerably higher use of [?] amongst the old speakers is found in the female speech as compared with the male speech, suggesting that the role of innovators is played 129 by female speakers, this being confirmed by the Tyneside findings. The role of female speakers as innovators is given further credence by the fact that the young females in the sample are found to be virtually wholly responsible for the use of the innovatory form, [ht], in the data, which also follows the pattern found in

Tyneside and suggests a degree of convergence.

The variants which have a marginal or low level of use in the speech of the old groups appear to be disappearing in the speech of the young. As the marginal variants appear to drop out of use in the speech of the young, the variant pool is reduced. There is thus less variability in the speech of the young compared with the speech of the old, which at first sight may appear to be indicative of a levelling process. However, this is offset by the adoption of new variants into the variety, as demonstrated by the use of [ht] amongst the young speakers of the sample.

In phonological terms, the decline noted in the marginal variants is also interesting. The decreasing use of 0 in the speech of the young combined with the marked increase in the use o f [?] suggests that [?] is not a stage on the way to elision of /t/. If it were the case that speakers were applying a synchronic rule of weakening in their use of variants of (t), an increase in use of 0 would be expected. The increased use of [?], then, appears to be socially motivated.

Overall, then, the data appear sharply differentiated by both age and gender and in terms of discoursal as well as phonological variation. A particularly interesting finding is the variation in evidence between the four groups of young speakers. Marked differences in the speech of the young of the sample are revealed both between the two genders and between the two age groups. The female adolescents were found to be almost wholly responsible for the use of [ht] in the data, whilst the young adult males are those who make considerable use of the discourse strategy of sentence tags. As well as investigating general trends across age in the data which may be indicative of change in progress, then, whether such fine-grained age differentiations are revealed in the other variables under consideration is of interest in the study. We turn now to an analysis of the

131

4.2 (p t k) Glottalling and glottalisation

4.2.1 Variants under consideration

All three variables, (p), (t) and (k), can be realised with glottal stop variants, globalised variants or fully released variants. However, a distinction must be drawn between glottalling (the realisation of (p), (t) or (k) as [?]) and

glottalisation, also termed glottal reinforcement. This refers to the realisation of (p), (t) and (k) as what is usually transcribed as a double articulation, [?p], [?t],

[Ik] or [p?], ft?], [k?].

Glottalisation has been variously described as, ‘glottal masking of the oral plosive burst’ (Wells 1982: 374), or ‘oral closure reinforced by a glottal closure' (Gimson 1989: 159). Giegerich (1992: 220) describes the phenomenon thus: ‘in syllable final voiceless stops the bilabial, alveolar or velar closure is accompanied - often slightly preceded - by glottal closure, so that a glottal stop [?] is co-articulated with the [p t k] articulation’. Recent acoustic analysis of glottal and globalised variants of (t) has revealed a more complex picture, however. This is both in terms of description of the phenomena and in terms of the fact that variation found in acoustic analysis of the variants correlates with particular speaker variables (for further discussion see Docherty and Foulkes 1999). In terms of a description of glottalisation, acoustic analysis has led to the suggestion of [d] being used to transcribe globalised (t) (Foulkes, Docherty and

Watt 1999: 7), given the predominance of full or partial voicing and the lack of visual evidence of a full voiceless glottal closure in spectrograms of the tokens analysed. Despite the somewhat arbitrary nature of the choice of terminology and transcription, then, for the purposes of this study the term glottalisation will be used to refer to the phenomenon, with tokens transcribed as [?p], [?t] and [?k]. 132

4.2.2 Geographical and social distribution of variants under consideration

Customarily cited as a feature of London English, it is problematic to determine the origins of [?] or precisely how old the [?] realisation of voiceless stops is, as no alphabetic letter exists for the glottal stop. Old texts, therefore, offer little in the way of evidence. A ‘glottal catch’ was first recorded as a feature of the west of Scotland, rather than London, however, by Alexander M. Bell (1860). According to Andresen (1968), who offers a detailed history of (t)-glottalling, it was noted by Sweet (1908) as occurring in ‘some North English and Scotch dialects’ before it was noted as a feature of London pronunciation (by Jones 1909). Indeed, using evidence from the Survey of English Dialects (Orton et al. 1962-71), Trudgill (1999a: 136) argues that glottalling of intervocalic (t) may have spread from the Norwich area to London, rather than from London to Norwich as often assumed. Others suggest the possibility of Glasgow being the origin of the feature (Macafee 1997: 528). Nonetheless, the use of the glottal stop as a variant of (t) is repeatedly cited as a Cockney characteristic (Wells 1982), or as a feature of what is termed ‘Estuary English’ (Rosewarne 1984, 1994; Coggle 1993).

Despite its general association with London English, however, use of the glottal stop for (t) is now considerably more widely distributed in urban areas of Britain. Its association with the seemingly influential London English is argued to be a contributory factor in its spread: ‘[i]t is certainly plausible to suppose that one of the principal factors contributing to the apparently recent geographical spread of T Glottalling is the influence of London English, where it is indeed very common’ (Wells 1982: 323). The spread of (t)-glottalling is seen as a fairly recent phenomenon, the extent of the dissemination leading Docherty et al. (1997: 282) to maintain that ‘[t]here is no doubt that glottalling (and the realisation of l\J as [?]) has become increasingly established in British English during the twentieth century’, and leading Trudgill (1999a: 136) to call it ‘one of the most dramatic, widespread and rapid changes to have occurred in British English in recent times’. Traditionally a stigmatised feature, it is ‘widely perceived as a stereotype of urban British speech’ (Milroy, Milroy and Hartley 133

1994: 5), and in Britain, according to Wells (1982:35), ‘is widely regarded as ugly and also a lazy sound’.

Sociolinguistic profiling of the glottal stop has been undertaken by researchers in many areas (Trudgill 1974; Romaine 1975; Newbrook 1986; Mees 1987). According to the recent studies detailed in Urban Voices (Foulkes and Docherty 1999b), such is the increase of glottal replacement of (t) that it is reported as being found in Newcastle, Derby, Sheffield, West Midlands, Milton Keynes, Reading, Hull, Norwich, South London, Cardiff, Glasgow, Edinburgh and (London)Derry. Use of [?] for (t) has also recently been reported in the speech of the young as far north as Huntly in the North East of Scotland (Marshall 2001).

With emphasis on the spread of (t)-glottalling in recent work, glottal replacement of the other two voiceless stops (p) and (k) in intervocalic word-medial positions appears under-researched in comparison. Early references to glottal replacement of (p) and (k), according to Andresen (1968: 25), are from Grant (1913) who mentions glottal replacement in Scotland, Hirst (1914) who states that glottal replacement of (k) is heard occasionally in Lancaster, and Matthews (1938) who states that ‘[?] replaces [t] and [k] between vowels, and that there is a growing tendency for it to replace [p]\ Little evidence in recent works seems available for the correlation of glottalling of word-medial (p) and (k) with particular speaker variables, however. Also, little seems to be known of the geographical distribution of glottal replacement of word internal (p) and (k) and whether it is spreading as is (t)-glottalling. Glottal replacement of (p) is considered a feature of London English, and has been noted in Edinburgh (Johnston 1985b) and Glasgow (Macafee 1983). It is also reported for Milton Keynes and Reading (although interestingly, for present purposes, not for the northern city of Hull) by Williams and Kerswill (1999).

Glottalisation of word medial (p), (t) and (k) is one of the most salient features of Tyneside English (Wells 1982, Milroy, Milroy and Hartley 1994, Docherty et al. 1997, Watt and Milroy 1999). It is also found in Durham (Kerswill 1987). However, Docherty et al. (1997: 306) claim that glottalisation, particularly of 134

intervocalic (p) and (k), may be recessive, arguing that it is characteristic of Tyneside male speech and more conservative rural varieties as found in south­ west Scotland and Northern Ireland. Although glottalisation is not recorded as a feature of Teesside English in the SED data, neither is it recorded as a feature of Tyneside English (Orton, Sanderson & Widdowson 1978: Map PH 239), it is taken, nevertheless, as a localised feature of the North East of England and therefore the extent of its use in MbE will be of interest.

4.2.3 Particulars of the data and areas of interest

In the conversational Middlesbrough data, 30 tokens were sought for all speakers for all three variables (p), (t) and (k). The chosen environment was word-medial, intervocalic, for example, paper, butter, wicked, where the token did not immediately precede a stressed vowel, as aspiration is considered to be strongest in the initial position of stressed syllables (Giegerich 1992: 219). The word- medial, intervocalic environment is cited as being the most stigmatised position for the glottal stop (Fabricius 2000, Wells 1982). Instances of (p), (t) and (k) preceding a syllabic /l/ or Ini were also included, as in people, happen, bottle, Brotton, tickle, taken. In citation form three tokens of word-medial (p), eight tokens of word-medial (t) and four tokens of word-medial (k) were included.

The Middlesbrough data are compared closely with the recent findings of the study of Tyneside English carried out by Docherty, et al. (1997). The extent of glottalisation in MbE as compared with that found in Newcastle English will be of interest. Additionally, as use of word-medial [?] is a feature which is spreading dramatically throughout BrE, a comparable amount of glottalling might be expected in both localities. This is particularly in view of the fact that the spread of |?] for (t) is argued not to originate necessarily from the South of

England, London and Glasgow/Edinburgh being cited as ‘dual epicentres' (Kerswill and Williams 1997: 245).

We turn now to the findings. Before correlating the data with particular speaker groups, the overall distribution of variants of (p), (t) and (k) are considered. We 135

then turn to an analysis of (p) broken down by age, gender, then age and gender. We proceed with examination of the distribution of variants of (p) in word-list style. We then follow the same pattern for (t) and finally (k).

4.2.4 Overall distributions of variants of (p), (t) and (k)

Although glottalisation affects (p), (t) and (k), in the Tyneside findings, glottal replacement, [?], was essentially found to affect only (t) (a 1% use of [?] for (p)

and 0% for (k) was recorded in Tyneside). In the Middlesbrough data this is not the case, as Table 4.6 below reveals.

Table 4.6: Percentage use of released, glottalised and glottallcd variants of (p t k) (conversational style, all speakers)

Total (word-medial) [p] [t] [k] [?pl [?t] [?k] m n. n. % n. % n. % (P) 899 410 45.6 454 50.5 35 3.9 (t) 947 222 23.4 158 16.7 567 59.9 (k) 833 564 67.7 203 24.4 66 7.9

If we consider use of the glottalled variants, at the right of Table 4.6, an overwhelming difference in the use of the glottal stop for (t) as opposed to its use for (p) or (k) is revealed. Unlike the findings for Tyneside, 66 instances of glottal replacement of (k) and 35 instances of glottal replacement of (p) are in evidence in the Middlesbrough data. Table 4.6 indicates that (k) is twice as likely to be realised as [?] than is (p). Although glottal replacement of (p) or (k) has been identified in such places as Milton Keynes, Reading, London, Glasgow and Edinburgh, as noted, little evidence is available to indicate which speakers use this particular variant for (p) and (k), or for which variable is more likely to be realised as [?]. In the Middlesbrough data, however, the ordering would seem to

be (t) > (k) > (p). 136

Table 4.6 also clearly shows that [?] is the preferred realisation of word-medial

(t) for the whole sample of speakers, with over half of all tokens being accounted for by [?]. As the Tyneside data revealed only a 16% female and an 8% male use of I?] for (t) (Docherty et al. 1997: 301), the extent of the difference between the

Middlesbrough data and the Newcastle findings is perhaps surprising given that the spread of [?] for (t) is not necessarily from the South of England, but may also be spreading from Scotland, and also given that forms are thought to descend an urban hierarchy of large city to city to large town (Britain 2000: 62), as noted on section 2.2.1.

Considering overall scores for the whole sample of speakers for use of the globalised variants [?p], [?t], [?k], marked differences between the variables are also revealed. Table 4.6 indicates that (p) is far more likely to be realised as a globalised variant than (k) or (t), with [?t] being the least likely realisation of (t).

Although the globalised variant is the preferred variant of (p) in the Middlesbrough data, its use appears substantially lower than in Tyneside (87% male and 58% female). Nevertheless, the higher susceptibility of (p) than (t) or (k) to glottalisation follows the same pattern as that found in Tyneside (Docherty et al. 1997: 301).

As regards the use of the released variants for the individual phonological variables, as presented at the left of Table 4.6, again, differences between the three variables are uncovered, with a substantially larger number of released tokens for (k) in evidence than for (p) or (t). Without yet considering age and gender variation, then, marked variation has already been revealed in the use of the variants of the variables (p), (t) and (k). Scores for the whole sample of speakers show that (p) is more susceptible to glottalisation, (t) is more susceptible to glottalling and the released form is the most likely realisation of (k). 137

4.2.5 (p), conversational style

4.2.5.1 Generational variation

Figure 4.9 below shows generational differences in the distribution of variants of (p), using the combined scores of the 4 young speaker groups. Percentage scores, therefore, are given for older, middle and young speakers.

Figure 4.9: (p), generational differences

70

61.1 60 57.6

49 3 48.5 50

40 ■ released ggglottalised 30 (□glottal stop

20

10

0 4 old (N. 225) middle (N. 216) young (N. 458)

A considerable increase in the use of [?p] and a marked decrease in the use of [p] is revealed across age from old to young. [?p] accounts for almost two thirds of the tokens of the young speakers, a level of use which is higher than both the middle and the older speakers’ scores. This suggests a degree of convergence towards speech of further north. As well as demonstrating the highest level of use of [?p], the young speakers have the lowest level of use of [p], indicating a degree of divergence from the non-localised or standard form. Additionally, although [?] is found occasionally across age groups, there is an increase in its use amongst young speakers. This increase may indicate the increased adoption of a possible current innovation whose use will continue to spread. 138

4.2.5.2 Gender

Figure 4.10: (p)> Sender

male (N. 454) female (N. 445)

Significant gender differences are revealed in variant usage (p<0.0001), with male speakers overwhelmingly preferring the localised [?p] and females showing a preference for the non-localised [p]. Overall, this was the finding for Tyneside, although differences of degree exist between the two localities. The percentage use of [?p] by the males is close to that shown by the males in the Tyneside study

(87%) (Docherty et al. 1997: 301). The females in the Middlesbrough data show a much lower level of use of [?p] compared to those in Tyneside (58%). In use of the localised variant, [?p], compared with the non-localised variant, [p], there is almost a complete mirroring in usage in terms of gender variation, with the localised variant associated with male speech and the non-localised variant associated with female speech.

Females demonstrate a higher level of use of [?] than males, however. The male speakers disfavour [?] virtually categorically, with only four tokens noted from

454. The females, on the other hand, show some level of use of [?], albeit to a small degree (31 tokens from 445). Whether there is an age and gender correlation in some of the trends revealed so far is now revealed. 139

4.2.5.3 Age and gender

Figure 4.11 presents the distribution of variants of (p) for male speakers by age.

Figure 4.11: (p), male speakers

old male middle young adolescent (N. 117) male (N. adultmale male (N. 103) (N. 120) 114)

Across age little consistent variability is in evidence in the male speech, with [ ?p] revealed to be overwhelmingly preferred by male speakers of all age groups.

A decline is in evidence in use of [?p] in the middle group, where the lowest level of use is demonstrated. Interestingly, the level of use of [?p] also falls in the speech of the adolescents compared with the young adult males, who show near categorical use of the localised variant. An undulating pattern is revealed across age, in which the distribution of variants in the speech of the old males is very similar to the speech of the young adult males, and the middle males reveal a very similar pattern of use to that of the adolescent males. There is some evidence of [?] use in the male speech. However, this amounts to one token from the middle group and three tokens recorded from the young males. Whether more variability is revealed in the female speech is seen in Figure 4.12 below. 140

Figure 4.12: (p), female speakers

old female middle young adolescent (N. 108) female (N adult female (N. 113) female (N. 112) 112)

As well as demonstrating the variation between male and female speech in the data, Figure 4.12 reveals considerably more variation in the speech of the females across age than was found in the male data. The older female speakers and the middle female speakers demonstrate virtually identical patterns of use, with a near categorical use of non-localised [p] demonstrated compared with marginal use of [?p] and [?]. Use of [p] then declines in the speech of the young adult females and we note an increase in use of both [?p] and [?]. The speech of the adolescent females is significantly different, however. Use of [p] falls to almost half the level recorded in the speech of the young adult females revealing a significant difference (p<0.010). Likewise, a significant increase in use of [?p] is observed (p<0.014), with the localised variant becoming the preferred variant of the adolescent females. Use of [?p] in the speech of the adolescent females is significantly lower than the level of use revealed in the speech of their male counterparts (p<0.002). However, the level of use in the female adolescent speech stands in stark contrast to that found in the speech of the old, middle and young adult females, and would seem to indicate that the female adolescents in Middlesbrough are at the vanguard of the convergence of MbE with varieties further north. 141

Similarly, the young females appear to be responsible for use of [?] in the data.

Seven of the eight young female speakers of the sample show some use of [?], compared with three of the other eight female speakers and three of all 16 male speakers. Interestingly, from the tokens recorded as [?] realisations of (p), 94.3%

(33 out of 35) were in cases where the (p) preceded a syllabic /l/ or /n/, for example people.

The findings for (p), then, reveal marked gender variation, generational differences which suggest change in progress in MbE, and fine-grained age differences which show that the speakers at the vanguard of the convergent trend are female adolescents. The speech of the older speaker groups reinforces the assertion that male speakers use a higher level of localised forms and female speakers use a higher level of non-localised forms. The sudden increase in the young females’ use of [?p] however, which has developed over time from virtual rejection at 4.6% to the preferred variant at 48.2%, indicates that MbE is converging with varieties found further north in Tyneside, Wearside and Durham. Additionally, the increase in the use of [?] suggests that [?] as a variant of (p) is spreading. This feature may continue to increase amongst young speakers and may prove a fruitful line of future research. Before looking at the findings for (t), we turn to an examination of the distribution of variants of (p) in word-list style.

4.2.5.4 (p): citation forms

Three items were included in the word-list to ascertain stylistic variation in the use of the variants under consideration. Table 4.7 below reveals the overall distribution of variants in word-list style.

Tabic 4.7: Overall distribution of variants of (p), word-list style

[P] [?p] m Total n. % n. % n. % 96 74 77.1 21 21.9 1 1 142

Almost three-quarters of tokens in word-list style are realised with the non­ localised [p]. This is in contrast with less than half of the tokens in conversational style (see Table 4.6). Use of [?p] decreases considerably from

50.5% in conversational style to 21.9% in word-list style. Evidence suggests that the localised variant appears to be subject to style shifting. The distribution of variants in word-list and conversational style for each speaker group is detailed below in Table 4.8.

Table 4.8: Distribution of variants of (p), word-list style (WLS) and conversational style (CS)

[p] % [?p] % (?) %

WLS CS WLS CS WLS CS Om 75 11.1 21 88.9 0 0 Mm 91.7 27.2 8.3 71.8 0 1 YAm 0 7.5 100 91.7 0 0.8 Adm 91.7 21.9 8.3 76.3 0 1.8

O f 100 90.8 0 4.6 0 4.6 M f 100 92 0 6.2 0 1.8 Y A f 100 78.6 0 11.6 0 9.8 A df 58.3 40.2 33.3 48.2 8.3 11.6

The majority of speakers use [p] in word-list style, demonstrating the expected style shift to the non-localised variant in the more formal speech style. Three of the female groups use [p] categorically. Although some use of |7p] is in evidence amongst the male speakers, three of the male groups also overwhelmingly prefer |pj in word-list style. However, two groups stand out quite markedly in their use of variants in word-list style: the young adult males and the adolescent females.

Categorical use of |?p] is demonstrated by the young adult males, which is in complete contrast to all the other groups of speakers. The young adult males also reveal the highest level of use of [?pj of all speakers in conversational style. 143

Nevertheless, their complete lack of style shifting in a more formal style is surprising given the other groups’, particularly the other young males’, use of [p] in word-list reading. The other group of speakers of interest is the adolescent female group. Again considerable use is made of the localised |?p] in word-list style. This stands in stark contrast with the categorical use of [p] by all other female speakers of the sample. This is also seen in the light of the adolescent female speakers’ increased level of use of [?p] in conversational style, however.

To some extent, then, the expected style shift in word-list style is in evidence in the data. However, the speakers with an increased use of the localised variant appear less likely to style shift away from it in a more formal speech style. We turn next to the findings for word-medial (t), beginning with the overall distribution of variants in conversational style.

4.2.6 (t), conversational style

4.2.6.1 Generational variation

Figure 4.13 below presents the distribution of variants of (t) across age. Young speakers’ scores are combined, giving groups of old, middle and young speakers.

Figure 4.13: (t), generational differences

88.7 90

80

70

60

50 ■ released 43.5 42.1 41 4 B glottalised 38.7 40 □ glottal stop

30

20

10

0 old (N. 235) middle (N. 232) 144

A dramatic increase across age from old to young is revealed in the use of [?] for

(t). Use rises from 19.2% amongst older speakers to near categorical 88.7% amongst young speakers. Although in line with the rapid spread of the glottal stop in urban varieties of BrE, as noted, the extent of its use amongst young speakers in the sample and the rapidity of its increase are quite remarkable.

Coupled with the increase in the level of use of [?] is a marked decrease in the levels of use of [t] and [?t]. The non-localised [t] slips from being the preferred variant of both the older and the middle speakers to near categorical rejection in the speech of the young. The speed of the declining use of [t] can be seen by comparing the level of use amongst the middle speakers and the young speakers. Similarly, a substantial decrease is noted in the level of use of [?t], which falls to a very low level of use amongst young speakers. We turn now to look at how the variable of gender acts on the data for the variable (t).

4.2.6.2 Gender

Figure 4 .14 below presents variant usage of (t) by gender.

Figure 4.14: (t), gender

70

61 n a a

■ released aglottalised □ glottal stop

male (N. 480) female (N. 467)

Figure 4.14 reveals that [?] is by far the preferred variant of both males and females in the sample. Although the male and female scores for [?] are 145

comparable, it is the female speakers who show a slightly higher level of use of [?], contrary to the assumption that [?] is a male norm. Although the

Middlesbrough data indicate a higher [?] incidence amongst female speakers

than males (which was the pattern found in the Tyneside data), use of [?] for (t)

is far more widespread in Middlesbrough than in Tyneside. In Middlesbrough over half of the tokens of both male and female speakers are accounted for by [?]

compared with 16% of the female tokens and 8% of the male tokens in the Tyneside study. As regards the other two variants, the pattern of variation resembles that found in the data for (p). Incidence of the localised [?t] is

significantly higher amongst male speakers than females (p<0.002). However, it is markedly lower than the Tyneside male usage of 82% (recall that a comparable

male use was found for [?p]). The female speakers avoid [?t] virtually categorically, and a significant gender difference is found in the use of [t] (p<0.001), as females favour the non-localised [t] as an alternative to [?] for (t).

We turn now to a consideration of the interaction of age and gender on the data.

4.2.6.3 Age and gender

Figure 4.15: (t), male speakers

100 95.8

90 80 80 75.8

70

60 ■ released ge 50 47.5 ■ glottalised □ glottal stop 40 27.5 30 25

20 14.2 10 1.7 2.5

old male ■middle young adolescent (N. 120) male (N. adult male male (N. 120) (N. 120) 120)

The steady and sharp increase in the level of use of [?] is revealed. However, we note a significant decline in use in the speech of the adolescent males as 146 compared with the near categorical use found amongst the young adult males

(p<0.036). The corresponding decline in the level of use of the localised [?t] mirrors the rise in [?], as it reaches its lowest level of use in the speech of the young adult males and then shows an increase in the speech of the adolescents. Use of [t] amongst male speakers is low across age. The older males show some use of [t], preferring it slightly over [?] which is the least preferred of the three variants in the old male group. Use of [t] rises to its highest point in the speech of the middle males, as did [p]. It then declines sharply in the speech of the young males. Whether the same patterns of variation are found in female speech is revealed in Figure 4.16 below.

Figure 4.18: (t), female speakers

100

old female middle young adolescent (N. 115) female (N. adult female (N. 112) female (N. 120) 120)

As was found in the male speech, dramatic variation is in evidence in the speech of the females of the sample across age. The increase in the use of [?] resembles that found in the male data. The most marked increase is found between the middle females and the young adult females, as was found in the male data. Use then increases further in the speech of the adolescent females, unlike the pattern found in the male data. The old female speakers also demonstrate a substantial level of use of [?]. Interestingly, by far the majority of the realisations of [?] amongst old speakers, both female and male, were found before syllabic /l/ or Ini. 82.2% (37 out of 45 tokens) were found in these environments in the speech of 147

the old. This was also the case in the use of [?] for (p), as noted. This may

indicate that [?] for (p) will spread to other intervocalic environments, following

the pattern of [?] for (t).

A marked decline is observed in the level of use of [t] amongst female speakers, as it was amongst male speakers. Again, the steepest drop comes between the middle speakers and the young adult speakers. The young females also show a sudden, albeit slight, increase in use of the localised [?t]. This is interesting,

again, in terms of the convergence of MbE with varieties found further north. Although the incidence of [?t] amongst young females is low, it is seen in

comparison with complete rejection by older females.

In the distribution of variants of (t) amongst female speakers, older and middle females show similar patterns, with evidence of a slight increase in [?] and

decrease in [t]. The older and middle female patterns are not only very different from those demonstrated by male speakers of the same ages, but also from those demonstrated by the young females. The young females’ speech is much closer to the young males’ speech in the use of variants of (t). This indicates a levelling of the gender variation found in the data thus far. Use of [?] is so high amongst young speakers that it seem doubtful whether use of the localised [?t], which shows a slight increase in usage, will make any real inroads into female speech, as seems to be happening with [?p].

With the variants of (t), then, again we have evidence of gender differences, generational differences suggesting change in progress, and fine-grained age differences. Older and middle speakers demonstrate similar patterns to those found in (p); male speakers prefer the localised [?t], whilst female speakers reveal a preference for the non-localised ft]. The young speakers, however, have converged on a variant, [?], and thus erased the gender differences. Every speaker in the sample demonstrates some use of [?]. With virtual categorical use amongst young speakers, both male and female, use of [?] for (t) seems to have 148 eradicated both the localised [?t] and the non-localised [t] almost completely.

This finding is also interesting in view of the much lower usage of [?] found in the Tyneside data. Indeed, such is the extent of young speakers’ use of (t)- glottalling in Middlesbrough that use is higher than that found not only in the northern city of Hull, but also in the South Eastern town of Milton Keynes, as found in the recent work by Williams and Kerswill (1999). Before examining the distribution of variants of (k), the findings for (t) in word-list style will be considered.

4.2.6.4 (t): citation forms

Eight instances of word-medial (t) were included in the word-list. Table 4.9 reveals the overall distribution of variants in word-list style.

Table 4.9: Overall distribution of variants of (t), word-list style

[t] |?t] I?1 Total n. % n. % n. % 254 163 64.2 36 14.2 55 21.6

We see a marked difference in use of [t] and [?] in word-list style as compared with conversational style (see Table 4.6). In word-list style, almost two thirds of the tokens are realised as [t], compared with 23.4% in conversational style. Similarly, [?] accounts for 21.6% of tokens in word-list style, unlike in conversational style where it accounts for almost 60% of the tokens. Whether individual speaker groups demonstrate a shift in style is revealed in Table 4.10 below. 149

Table 4.10: Distribution of variants of (t), word-list style (WLS) and conversational style (CS)

[t]% [?t]% [?]%

WLS CSWLS CSWLSCS Om 78.1 12.5 18.8 75.8 3.1 11.7 Mm 61.3 25 35.5 27.5 3.2 47.5 YAm 9.4 1.7 25 2.5 65.6 95.8 Adm 45.2 5.8 32.2 14.2 22.6 80

O f 100 73 0 0 0 27 M f 96.9 63.4 0 1.8 3.1 34.8 Y A f 65.6 6.7 0 5 34.4 88.3 A df 56.2 4.2 3.1 5 40.7 90.8

Considerably more variation is in evidence in word-list style for (t) than was found in the word-list data for (p). Despite this, the non-localised [t] is still the preferred variant for all but one speaker group. As with (p) in word-list style, the older females show categorical use of [t]. The middle females also demonstrate virtually categorical use of [t], with only one token realised as [?] in word-list style. Given the stigma attached to word-medial, intervocalic [?] for (t), the amount of [?] usage in the formal word-list style is surprisingly high, especially amongst young speakers. Use of [?] is revealed in all groups except the old female group. The young female speakers show a high level of use of [?] in word-list style, with the highest use found in the speech of the female adolescents, who also revealed the highest female level of use of [?| in conversational style. The female adolescents are also the only females to show some use of the localised [?t] in word-list style. All male groups show some use of [?t], however. Additionally, all male groups show some use of [?], although in the old male group and the middle male group this amounts to one token only. This would seem to indicate that the male speakers style shift less than the females. As in the word-list data for (p), the young adult males stand out amongst 150 the other male groups, as do the adolescent females amongst the other female groups. The young adult males have by far the lowest score for [t] and the highest score for [?]. However, at a near categorical 95.8%, the young adult males demonstrated the highest level of use of [?] in conversational style in the data. The same pattern as found for (p) thus emerges. The speakers with the highest level of use of a localised or stigmatised variant in conversational style are the speakers who are less inclined to use non-localised or prestige forms in word-list style, these speakers being the young adult males and the adolescent females. We turn now to the distribution of variants of word-medial (k).

4.2.7: (k), conversational style

4.2.7.1: Generational variation

Figure 4.17 below presents variant usage of (k) across age. Again, generational differences are analysed and the young speakers’ scores are combined.

Figure 4.17: (k), generational differences

old (N. 211) middle (N.224) young (N. 398)

A certain amount of variation across age from old to young is revealed in Figure 4.17. However the variation is not as marked in (k) as was seen with (p) and (t). The most striking thing to notice is the overall preference for the non-localised |k|. All age groups show a substantially higher incidence of (k| than of |?k] or 151

[?]. Over two thirds of the tokens in all age groups are accounted for by [kj, with the older speakers and the young speakers showing very similar levels of use. The middle group demonstrates the highest level of use of [k] at 75%. A similar pattern in use of [?k] is revealed as that found in the use of [?p]. Old and young speakers reveal similar scores, but a decline is noted in the speech of the middle group. The highest incidence of [?k] is found in the speech of the young of the sample, as found for [?p]. This suggests a degree of convergence with varieties further north. Unlike the findings for (p) and (t), use of [?] for (k) shows a slight, though surprising, decline amongst young speakers, with older and middle speakers revealing a virtually equal level of use. Across age groups, then, incidence of [?] for (k) is small but persistent1.

Overall, a similar pattern in the distribution of variants of (k) is demonstrated in the speech of the old and the young. The middle group deviates from the pattern somewhat by showing a higher incidence of [k] and a lower incidence of [?k].

Such variation corresponds with the U-curve effect, detailed in section 3.1.3. We turn next to a consideration of the distribution of variants by gender.

4.2.7.2 Gender

Figure 4.18 below presents gender variation in the variant usage of (k).

1 Additionally, three of the young speakers (one male and two females) demonstrated voiceless velar fricative realisations of (k), such as ['naxad] knackered ftw D xaz] twockers fb exi:] Becky. This is a feature associated with Merseyside (Newbrook 1999, Wells 1982) and whether its use will spread in Middlesbrough, or whether its emergence is simply a case of individual idiosyncrasies, is impossible to say without further tracking of this variant. 152

Figure 4.18: (k), gender

90 ___ 86 7

male (N. 435) female (N. 398)

Both males and females use [k] in the majority of tokens. However, a significantly higher level of use is found by the female speakers (p<0.0001). A significant gender difference is also noted in use of [?k] (p<0.000l), where the same gender-correlated variation as that found for (p) and (t) is revealed. Males demonstrate a considerably higher level of use of [?k] than females, from whom only 10 tokens are recorded from 398. The level of use of [?k] by both males and females is lower than that found in Tyneside, however, where an 82% male use and a 37% female used is revealed (Docherty et al. 1997: 301).

Female speakers demonstrate more than twice the incidence of [?] for (k) than do the males. Interestingly, the females prefer [?] to [7k] as a realisation o f (k), unlike the findings for Tyneside where the females showed no use at all of [?] for

(k). Although [?] is the males’ least preferred variant it is still in evidence, as 23 tokens were recorded, unlike in Tyneside where, again, no tokens were in evidence. Whether significant patterns emerge when age and gender are both considered is now revealed. 153

4.2.7.3 Age and gender

Figure 4.19 presents the distribution of variants of (k) in male speech.

Figure 4.19: (k), male speakers

old male middle young adolescent (N. 101) m ale(N. adult male m ale(N. 107) (N. 120) 107)

The older males show a substantial preference for [?k] (although a much lower use of [?k] is revealed compared with the incidence of [Vp] and [?t] in the speech of the old males). The old males’ use of [k] is also fairly high, with over a third of the tokens accounted for by this variant. This is also seen in comparison with an 11.1% use of [p] and a 12.5% use of [t]. Also some use of [?] is in evidence amongst old males, at least one token being recorded from each speaker. The middle male speakers show a considerably different pattern from that of the older males, [k] accounts for over two thirds of the tokens from the middle males. Use of [?k| drops considerably, and again is seen in comparison to a 71.8% use of

["Vp]. [?] is virtually categorically rejected, with only one token recorded from the middle males. The young adult males demonstrate a pattern which is remarkably similar to that of the older males, as was found for (p). An increase in the use of [?] is noted, however, which is found in the speech of all four young adult males

in the sample. Again, we see a drop in use of the localised variant and an increased use in the non-localised variant in the speech of the adolescent males, which results in the adolescent males demonstrating a pattern which is very 154 similar to that found for the middle males. Thus we see exactly the same undulating pattern that was in evidence in the male data for (p). Figure 4.20 below presents the female data.

Figure 4.20: (k), female speakers

100 , 94.8 90 i 86 4

■ released B glottalised □ glottal stop

13.6 0

old female middle young adolescent (N. 110) female (N. adult female (N. 117) female (N. 94) 77)

Little variation is in evidence in the speech of the females. In all age groups by far the preferred variant is [k], which never falls below 80%. Interestingly, in this instance we find the middle speakers demonstrating the lowest level of use of the non-localised variant. However, use is still high and the differences slight. Unusually also, these results represent the only time that young speakers have been found to have the highest incidence of the released variant, with four of the eight young female speakers demonstrating categorical use of [k]. Overall, use of [?] for (k) is higher in female speech than in that of the males, as was the case for

(p) and (t). Unlike the findings for (p) and (t), however, the incidence of [?] amongst young speakers declines compared with that found amongst middle and old female speakers. The young females’ use of [?] is not only considerably lower than the older and middle females’, but is also lower than the young males' use. The middle females, unusually, reveal the highest level of use of [?], with every member of the group demonstrating some use. Again, as was found with (p) and the older speakers’ use of (t) variants, there appears to be a connection between use of [?] before syllabic Ini and /!/. In the case of the female speakers, 155

42 of the 43 [?] tokens recorded precede a syllabic /n/. Similarly, in the male findings, 21 of the 23 [?] tokens preceded a syllabic /n/. In the female data for

(k), there is the now familiar pattern of an increase in the use of the globalised variant amongst young speakers, which was found in both (p) and (t). Use is slight when compared with male use, but is seen against its categorical rejection by the older females. The increase in use is seen in the light of convergent linguistic trends in the North East of England to be discussed further in Chapter

6.

Overall, then, not as much variation is in evidence in the data for (k) as for (p) and (t), yet a similar pattern of an increase in the use of the local globalised variant has emerged. Despite the differences found between the two young groups of speakers, the combined young group has the highest incidence of [?k] in the sample, suggesting again the convergence of MbE with varieties further north. Overall, [k] was the preferred variant for both male and female speakers. Despite this, the same pattern of gender variation was revealed in the data as was found in variant usage of (p) and (t). The female speakers overwhelmingly prefer the non-localised variant, whereas the male speakers show a high incidence of the localised variant. Glottal stop realisations were most favoured by the females, which was also found in the data for (p) and (t). A surprising drop in use of [?] was noted amongst young speakers, however. Whether its use will decline further and whether use of [x] will spread remain to be seen and is another fruitful area for further research.

4.2.7.4 (k): citation forms

Four instances of word-medial (k) were included in the word-list. Table 4.11 below presents the overall distribution of variants. 156

Table 4.11: Overall distribution of variants of (k), word-list style

[k] [?k] I?) Total n. % n. % n. % 128 115 89.8 13 10.2 0 0

Use of [k] is near categorical in word-list style and is considerably higher than the incidence in conversational style (67.7%). Likewise, use of [?k] is lower in word-list style, suggesting a degree of style shifting. Whether this is the case for all speaker groups is revealed in Table 4.12 below.

Table 4.12: Distribution of variants of (t), word-list style (WLS) and conversational style (CS)

[k]% |?k]% (?]%

WLSCSWLS CS WLSCS Om 93.7 39.6 6.3 55.4 0 5 Mm 100 67.3 0 31.8 0 0.9 YAm 37.5 35 62.5 55.8 0 9.2 Ad m 100 60.8 0 33.6 0 5.6

O f 100 86.4 0 0 0 13.6 M f 100 82.1 0 0.8 0 17.1 YAf 100 94.8 0 2.6 0 2.6 Adf 87.5 86.2 12.5 7.4 0 6.4

Considerably less variation is revealed in the word-list data for (k) than for (p) and particularly for (t). This was found to be true of the conversational data also, however. In the word-list data, five of the eight speaker groups demonstrate categorical use of fk]. This includes, surprisingly perhaps, the adolescent male group. However, a drop was revealed in the use of [?k] in the speech of the adolescent males as compared with the young adult males in conversational style 157 also. Use of [?k] amongst the old male speakers amounts to one token only. As was found in data for (p) and (t), the pattern emerges in which two groups stand out as different from the others. These groups are the young adult males and the adolescent females.

The young adult males reveal a high incidence of [?k] compared with all the other speakers. All four members of this group showed some use of this variant in word-list style. Again, the young adult males showed the highest incidence of the glottalised variant of all speakers in the conversational data. As the young adult males’ use of [?k] in conversational style was virtually the same as the old males’ use, the young adult males’ score in word-list style suggests that they are less inclined to shift to a more formal speech style in word list reading than are other speakers, for example, the old males. The adolescent females also show some use of [?k] in word-list style. This amounts to two tokens, however.

Nonetheless, again, the female adolescents demonstrate the highest female use of [?k] in conversational style.

4.2.8 (p t k): summary

The variation found in the distribution of variants of (p), (t) and (k) not only has interesting implications at the phonological level and the sociolinguistic and stylistic levels, but also in terms of convergent and divergent linguistic trends.

At the phonological level, the data presented would seem to reinforce one of the conclusions drawn by the Tyneside study (Docherty et al. 1997), that the glottal and glottalised variants represent different choices available to the speaker which are not necessarily ranged on a continuum of increasing lenition. 1'he female use of the released variant as an alternative to the glottalled variant of (t) is not linear, there being a rejection of the intermediary glottalised form. Use of these variants, therefore, involves choices which are available and systematically preferred and dispreferred by speakers. Additionally, by analysing use of the glottalled and the glottalised variants of (p t k), findings reveal the higher susceptibilities of the individual phonological variables to glottalling and glottalisation. The pattern 158 found is the same as that found in the Tyneside data. That is, /p/ is the most susceptible of the three stops to glottalisation and /t/ is the most susceptible of the three stops to glottalling.

On the sociolinguistic level, in terms of gender variation, substantial differences were revealed in the use of the globalised variants and the released variants. The patterns found to some extent reinforce the idea that females prefer non-localised variants and males prefer localised variants. However, for all three stops the females demonstrated a higher use of the glottal stop than the male speakers. Although the glottal stop is not a localised variant, it is a stigmatised variant, and the increased female use of [?] is in line with many recent studies and contrary to the belief that [?] is a WC male norm. The age-correlated variation in the data was also interesting. The variable for which the most variation across age was revealed was (t). The increase in the use of [?] for (t) amongst young speakers reflects the dramatic and widespread increase of this variant throughout Britain. As it is used almost categorically amongst young speakers, it has not only levelled out the gender variation found in the other two stops but has also virtually levelled out the other two variants completely. Given the extent of its limited use in Tyneside, use of [?] amongst the young speakers of Middlesbrough is surprisingly high.

In terms of convergent and divergent trends, a significant finding has been that the combined group of young speakers showed the highest level of use of [?p| and [?k] of all speakers. This increase, and particularly the young females’ sudden use of globalised stops, is seen as evidence of the convergence of MbE with varieties from further north in Tyneside, Wearside and Durham. Interestingly, in the pilot study of Middlesbrough (Llamas 1998) which examined data from young speakers only, the low use of the globalised variants amongst young females was taken as evidence of the levelling out of the localised variant by the female speakers. On the contrary, the evidence presented here suggests that globalised stops are not recessive in MbE as they are in Newcastle and as was suggested by the results of the pilot study, but are, in fact, increasing in the speech of the young, thus producing a degree of convergence of 159

MbE with varieties further north. Furthermore, rather than leading a levelling tendency, the young females, particularly the adolescents, are at the vanguard of this convergent trend. 160

4.3 (r)

4.3.1 Variants of British English (r)

According to Gimson (1989: 209), there are more phonetic variants of the /r/ phoneme than of any other English consonant. Variants of BrE (r) have recently been discussed in detail by Foulkes and Docherty (2000, in press). Much of the background material to this section is thus derived from the aforementioned recent works. The alveolar approximant, [ j ] , is the most usual realisation of BrE

(r) (Hughes and Trudgill 1996: 41). Other dialectal or phonologically restricted variants of (r) found in BrE include the alveolar tap [r], the alveolar trill [r], the retroflex approximant [.j] and the uvular fricative [k]. Despite its omission from the list of rhotics proposed by Ladefoged and Maddiesen (1996: 245), and its omission as a variant in the study of the phonetics of rhotics by Lindau (1985), the labio-dental approximant [u]1 is also a variant of BrE (r). It has been argued that labialisation of (r) is the most recent stage in a progressive weakening of (r) in BrE (Murray 1998, Trudgill 1988, 1999b), with Trudgill (1999b) describing the history of lenition in the realisation of BrE (r) as [r > r > 4,> J > uj.

Despite being acknowledged as a variant of BrE (r), the labio-dental approximant has often been dismissed as a feature of infantilism, defective speech or upper- class affectation. The standard variant, the alveolar approximant, is a complex articulation which is late to develop in children (Cruttenden 1994: 189). As |u | is seen as lacking the articulatory complexities of [ j ] , it is often associated with children’s speech. Gimson, for example, notes that ‘such a pronunciation, once a fashionable affectation (e.g. in the late eighteenth and early nineteenth centuries) is now only heard amongst children who have not yet acquired [ j ] . ’ [ u ] has also been described as a ‘defective /r/’ (O’Connor 1973: 49) or ‘commonly substituted for [j] in defective speech’ (Gimson 1989: 208). Additionally |u].

1 The symbol [u] is traditionally used to refer to labialised (r) and is the symbol used in the present work. However, it is argued that the symbol [|i] may be equally appropriate (Foulkes and Docherty 2000, in press). 161 also known as ‘public school /r/’, is associated with ‘upper-class affectation’ (Wells 1982: 282). In literature, the use of w for r in speech has often been used to denote the labial (r) associated with upper-class or affected speech. For example, Dickens and Orwell both use w for r in the speech of effeminate or effete characters (see further Foulkes and Docherty in press). Despite the association of [u] with upper class speech, however, Wells (1982: 282) anticipates the spread of [u], stating ‘I am not convinced that it is nowadays found more frequently among upper-class speakers than among those of other social classes’. Despite the perception of [u] as a somehow defective (r), then, [u] must be considered a variant of BrE (r), the distribution and spread of which is considered in the following section.

4.3.2 Geographical and social distribution of variants under consideration

Three variants of (r) are under consideration in the present study: the alveolar tap [r], the alveolar approximant [j] and the labio-dental approximant [u]. According to Wells (1982: 282), the tapped variant is typical of some varieties of Upper- crust RP (U-RP) and is found intervocalically after a stressed vowel and also in certain consonant clusters. The use of [r], according to Wells, may contribute to the ‘crisp, clipped effect which many claim to detect in U-RP’ (1982: 282). Hughes and Trudgill (1996: 41) note that [r] may occur in RP between vowels when the first vowel is stressed and following a dental fricative. O’Connor (1973: 150) also states that [r] ‘may occur in RP and other accents in words such as marry, borrow, where /r/ is preceded by a short, stressed vowel and followed by an unstressed one’.

In terms of regional varieties of BrE, the tapped variant is associated with Scotland, Wales and Northern England (Hughes and Trudgill 1996: 61). Cruttenden (1994: 188) argues that the tap is ‘common in colloquial forms of both Scottish English and some northern urban speech’. Wells (1982: 368) also claims that ‘ [r] seems quite widespread in the north of England as a rival to the usual post-alveolar approximant [j]’. Wells maintains that the exact geographical 162 spread of [r] is not known, neither are the exact conditions favouring the use of

[r], although he claims ‘I have the impression that the environments in which it occurs most readily are that of a preceding labial [...] a preceding dental [...] and intervocalically’ (1982: 368). In the BrE accent studies presented in Urban Voices (Foulkes and Docherty 1999b), the tapped variant is reported as being found in Newcastle, Derby, Sheffield, West Wirral, Sandwell, Cardiff, Glasgow and Edinburgh, with most studies specifying an intervocalic position as the most likely environment in which [r] may occur. With the exception of Hull English where, interestingly, it is not recorded as a feature (Williams and Kerswill 1999), the tapped variant is found in all the featured localities in the north of England, Scotland and Wales.

The alveolar approximant is found throughout BrE. As such it is a non-localised variant. With the exception of Dublin, in all locations featured in Urban Voices

(Foulkes and Docherty 1999b), [ j ] is described as the most usual or typical realisation of (r). It can thus be seen as the ‘mainstream’ variant of BrE (r).

Despite the stigmatisation of [u] deriving from its association with infantilism, defective speech and upper class affectation, perceptions of labial (r) appear to be changing and [u] is currently spreading rapidly throughout English English. Use of [u] is noted widely in the speech of media presenters and public figures in

Britain, such as sports personalities, pop stars, actors, television presenters and Members of Parliament (see further Foulkes and Docherty 2000, in press). Use of [u] has also recently been noted in several urban varieties of BrE. In his initial study of Norwich English in 1968, Trudgill (1974) found the use of [u] in the speech of only a few speakers. Although according to Trudgill (1988: 40) in retrospect ‘the beginnings of a linguistic change’ were in evidence in the 1968 fieldwork, use of [u] in data from 1968 was regarded as ‘an idiosyncratic feature in the speech of those who had it [...] simply a speech defect’. However, in his return to Norwich in 1983, Trudgill (1988) revealed that 33% of speakers bom between 1959 and 1973 used [u], with the younger generation employing [u] to ‘an astonishing degree’. Kerswill (1996) compared the speech of children and 163 caregivers in Milton Keynes and found a marked split between parents’ and children’s use of [ j ] and [ u ] , with parents favouring [ j ] and children favouring the [u] realisation. The [w] variant was also found in the speech of four year old children but not in the speech of older children, suggesting to Kerswill that children ‘are acquiring [u] as a dialect feature, whatever their earliest realization of /r/ might have been’ (1996: 190). [u] has also been noted in the speech of the young in the South East of England (Torgersen 1997), and Reading (Williams and Kerswill 1999).

Considering reports on the use of labial (r) in localities further north of London, we find that [u] is ‘a widespread feature of speakers from various parts of

Lancashire and Cheshire, particularly common amongst teenagers and younger children (Lodge 1984: 30). Indeed, with the exception of Sheffield, [u] is recorded as a feature of the young speakers of all the English urban varieties investigated in Urban Voices. As far north as Newcastle, use of [u] is recorded as

‘increasing in young speech’ (Watt and Milroy 1999: 30). Such is the spread of [u] then that Foulkes (1997) claims that ‘ [u] must now be recognised as one of the principal variants of English English 111'. However, it is not recorded as a feature in Scotland (Jane Stuart-Smith, personal communication), Wales or Ireland. Additionally, although [u] is recorded as a feature of Newcastle, in the recent study of Newcastle and Derby English, use of [u] was found to be considerably greater in the city of Derby, which lies in the Midlands, than in Newcastle, situated in the North East of England (see further Foulkes and Docherty 2000, in press, Docherty and Foulkes in press).

Although no direct evidence exists to confirm that [u] is a feature of London

English which is diffusing northwards, [u] is described as a feature used by ‘young speakers in the south’ (Hughes and Trudgill 1996: 61). It is recorded as the main variant of (r) for eight to ten year old London schoolchildren (Beaken 1971). Barltrop and Wolveridge (1980: 100) state that ‘[i]n Cockney speech, r is pronounced by turning both lips outwards’, suggesting a labial articulation. 164

Furthermore, Wright (1981: 135) states that ‘[although it seems chiefly an individual matter, Greater London does appear to have a larger percentage of w- for-r people than do other parts of the country’. Such evidence in the literature, coupled with the fact that labio-dental (r) is not used in Scotland and is used far less in Newcastle than in Derby, suggests that the spread of [u] may be interpreted as a feature spreading from a South Eastern epicentre. It also appears from the literature that [u] is a feature associated with young speakers and indeed may ‘serve the role of a generation indicator’ (Nolan and Oh 1996: 44).

4.3.3 Particulars of the data and areas of interest

In the conversational data, 30 tokens were sought for all speakers for the variable (r). Tokens were analysed in intervocalic positions. As MbE is non-rhotic, this included word-medial position, for example, carry, area, and word-initial position where (r) followed a vowel, for example, a real, to reach. Utterance initial instances of (r) were also included, as were instances of (r) following a vowel and a glottal stop, (r) in post-vocalic, word-final position will be examined in a future work on linking and intrusive III.

In citation form, in an intervocalic environment, three instances of word-medial (r) and one of word-initial post-vocalic (r) were included. Three other tokens of word-initial (r) are elicited in citation form. However, the word-list effect is such that, in reading, no pause is inserted between citation forms by the majority of informants. Word-initial tokens, therefore, cannot be taken as instances of utterance initial (r). For a direct comparison of the effect of style-shifting in word-list style compared with conversational style, the four tokens in which (r) appears in intervocalic position are analysed.

Of the three variants of (r) under consideration, [r] is considered the conservative or relatively localised variant. It is considered relatively localised as the tapped variant is found in northern England, Scotland and Wales and therefore the majority of the British Isles in geographical terms. However, as [ j ] is found throughout BrE, it is non-localised by comparison, [u] is the spreading variant 165 which appears to be diffusing rapidly from the South East of England to urban centres further removed from London. The three variants under consideration then, [r], [j] and [u], are considered in relation to the definitions of localised, non-localised, and spreading variants respectively.

Before analysing the data in terms of correlations with particular speaker variables, the overall distribution of the variants under consideration in the conversational data is considered.

4.3.4 Overall distribution of variants of (r)

Table 4.13: Overall distribution of variants of (r) (conversational style, all speakers)

Total [r] M [u] n. % n. % n. % 959 126 13.1 700 73 133 13.9

Across speakers the preferred variant by far is the non-localised [j ]. [ j ] realisations of (r) account for almost three quarters of the total number of tokens analysed from the conversational data. This is in line with general findings from urban vernaculars throughout BrE, which show [ j ] to be the most common realisation of (r), noted in section 4.3.2. Interestingly, the other two variants under consideration, [r] and [u], have a virtually equal distribution in the data, with a 13% use revealed for both variants. There exists therefore an almost exact balance in the use of the localised and spreading variants of (r) found in the conversational data overall. Whether use of the variants in conversational style is correlated with particular speaker variables is considered below. We turn first to an analysis of generational differences in the distribution of variants, before considering how the extra-linguistic variable of gender acts on the data. We then examine the data broken down by age and gender, before a consideration of the effects of style shifting on the distribution of variants in word-list style. 166

4.3.5 (r), conversational style

4.3.5.1 Generational variation

Figure 4.21 below shows the distribution of variants of (r) across age, using the combined scores of the young speaker groups.

Figure 4.21: (r), generational differences

100 r...... -...... 90.9

old (N. 240) middle (N. 240) young (N. 479)

The preferred variant for speakers across age is revealed to be the non-localised

[ j ] . Older speakers demonstrate the lowest level of use of [ j ]. The middle speakers reveal the highest level of use which is near categorical. The combined group of young speakers show a use of [ j ] which, although higher than the older speakers’ score, is considerably lower than the middle speakers’ usage. Use of [r] sees a substantial drop across age from old to young, falling to virtual rejection amongst young speakers. The sharpest decline in the level of use of [r] lies between the old speakers and the middle speakers, where use of [r] falls dramatically. Use of the [u] is essentially only in evidence amongst the young speakers. The appearance of [u] in the data from the young speakers of the sample suggests the sudden and substantial emergence of the spreading feature in MbE. No instances of [u] are recorded from the old group, and the middle group reveal a use, which amounts to two tokens from 240. In effect, therefore, use of 167

[u] is confined to the young speakers of the sample. The sudden use of [u] amongst young speakers is combined with a declining use of both [j ] and [r].

Whether a gender correlation exists in any of the trends revealed so far is considered below.

4.3.5.2 Gender

Figure 4.22: (r), gender

male (N. 479) female (N. 480)

Marked gender differences are in evidence, with male speakers demonstrating a significantly higher level of use of [r] than females (p<0.011), adding further support to the assertion that male speakers generally use a higher proportion of localised variants than do female speakers. The females exhibit a use of the non­ localised [j], which is significantly higher than the male use in the data

(p<0.016), again reinforcing the belief that female speakers in general use a higher degree of non-localised variants than do their male counterparts. Interestingly, the gender difference in the use of [u] is not as marked as the gender difference revealed in the use of the other two variants under consideration. Although the male speakers of the sample exhibit a higher level of use of [u] than the female speakers, the difference is comparatively small and is not statistically significant. 168

4.3.5.3 Age and gender

Figure 4.23: (r), male speakers

90 ...... 85.9

80

70 63.3 59.2 60 ■ tap 50 U alveolar approx. 40 □ labio-dental approx.

30

20

10 0 old male middle young adolescent (N. 120) male (N. adult male male (N. 120) (N. 119) 120)

A considerable amount of variability exists in the distribution of variants of (r) amongst male speakers. The two things which are immediately striking in Figure 4.23 are the marked decline in the use of [r] and the emergence and substantial

increase of [u] in the speech of the young in the sample. The preferred variant of

the old male speakers by a large majority is the localised [r]. Use of [r] accounts

for almost two thirds of the tokens recorded from this group, with the variant being used in substantial numbers by all old male speakers. The sharpest drop in use of [r] lies between the old male and the middle male groups. Use of [r] is still

present in the speech of three of the four members of the middle male group, however. Use of [r] continues to decline steadily, reaching virtual rejection

amongst the adolescents, where only two tokens are recorded.

By contrast, use of the non-localised [ j ] remains comparatively stable across age

from old to young. Although it accounts for only just over one third of the tokens recorded for the old male speaker group, use of [ j ] rises in the speech of the speakers of the middle male group, which again corresponds with the U-curve effect, [j] is the preferred variant of all speakers in the middle male group.

However, use of [j] declines in the speech of the young males. A substantial drop 169

between the middle group and the young adult group is noted. Use of [ j ] then declines further between two young groups of male speakers.

[u] is completely absent from the speech of the old males in the sample. It is effectively absent from the speech of the middle males also, as one token is recorded from 120. Seen in this context, the substantial emergence and increase in use of [u] amongst the young speakers is quite remarkable. The extent to which the trends revealed in the distribution of variants of (r) in the speech of males in the sample are duplicated in the speech of the females is considered below.

Figure 4.24: (r), female speakers

100 , ...... 95.9

90

80

70

60 ■ tap 5? 50 mu alveolar approx. □ labio-dental approx. 40

30 20 10 0 old female middle young adolescent (N. 120) female (N. adult female (N. 120) female (N. 120) 120) The trends noted in the conversational data from the male speakers of the sample are in evidence in the data from the female speakers, although to a lesser degree. As was found in the speech of the male speakers, a considerable decline from old to young is noted in the use of [f] in the female speech. The old female speakers exhibit the highest level of use of [r] of the female speakers, although only two of the four speakers of the group demonstrated a use of [r]. Interestingly, these two females were the two older informants from the group of older females (ages 69 and 80). Neither of the two 60 year old female speakers in the group showed any use of [r]. Although the old female group reveal a fairly substantial use of [r] compared with other female groups, the old female use of [r] is significantly lower than use by their male counterparts (p<0.036). The middle group of female 170 speakers shows little use of [r], with three tokens from 120 recorded, emphasising the considerable decline in the use of [r] between the older and the middle groups of female speakers. The two groups of young female speakers categorically avoid the localised [r].

The non-localised [j ] is by far the preferred variant of the female speakers of all ages, with 15 of the 16 female speakers in the sample revealing [j] to be their preferred variant. Although use of [j] is higher amongst females than males, use o f [j] shows the same patterning for both males and females. Use of [j ] rises to its highest point amongst the middle group of speakers, among whom it is used virtually categorically, again illustrating the U-curve effect. All four speakers in the middle group demonstrate a score of at least 90% for [j]. Use of [j ] decreases in the speech of the young adults and declines further in the speech of the adolescent females, as was the case in the speech of the adolescent males, where it reaches its lowest point of use.

Again, as with the findings for the male speakers, use of [u] appears suddenly and to a fairly substantial degree in the speech of the young females of the sample. Use of [u] is absent from the speech of the older females and virtually absent from the speech of the middle group of female speakers, where, as was found in the male data, one instance of [u] was recorded. Again, as in the male data, use of [u] increases in the speech of the adolescent females as compared with the young adult females by approximately 10%.

We turn now to an analysis of the distribution of variants of (r) in word-list style.

4.3.6 (r): citation forms

Four tokens of intervocalic (r) were included in citation form to ascertain stylistic variation in the use of the variants under consideration. Table 4.14 below presents the overall distribution of variants in word-list style. 171

Table 4.14: Overall distribution of variants of (r), word-list style

Total [r] M [u] n. % n. % n. % 128 15 11.7 96 75 17 13.3

The majority of the tokens are accounted for by [j] as was the case in the conversational data. Indeed, when analysing the overall scores for the distribution of variants of (r) in word-list style, the findings are remarkably close to those found for the overall scores for the distribution of variants of (r) in conversational style (see Table 4.13). This would suggest that use of the localised and the spreading variants of (r) are not subject to style shifting in the more overtly careful speech style employed in a word-list reading exercise. Further, this lack of style shifting would suggest that both the localised [r] and the spreading [u] are indicators of variation as opposed to markers. The extent to which this is true when the word-list data are broken down into the speaker groups is revealed in Table 4.15 below.

Table 4.15: Distribution of variants of (r), (WLS) and (CS)

[r] % [j] % [u] %

WLS CS WLSCS WLS CS Om 62.5 63.3 37.5 36.7 0 0 Mm 6.25 13.3 93.75 85.9 0 0.8 YAin 6.25 7.6 56.25 63 37.5 29.4 Adm 0 1.7 75 59.2 25 39.1

Of 18.75 15.8 81.25 84.2 0 0 Mf 0 3.3 100 95.9 0 0.8 YAf 0 0 81.25 85 18.75 15, Adf 0 0 75 74.2 25 25.8 172

The distribution of variants of (r) in word-list style corresponds very closely with that in conversational style. Remarkably, the scores are almost the same in most cases, with a difference of over 10% recorded for the adolescent males only. Such findings may lead us to believe that no effects of style shifting are observable in the data for (r). It is important to note that differences are discernible in the use of [u] amongst individual speakers, however, as the speakers with the greatest number of [u] realisations in conversational style from each group are the speakers demonstrating [u] usage in the word-list style. In terms of the use of [r], by far the highest level of use is exhibited by the old males in word-list style, as was found in the conversational data, with all the members of the group revealing some use of [r] in word-list style. The use of [f] in the old female group is accounted for by the speaker of the group who revealed the highest level of use of [r] in the conversational data.

4.3.7 (r): summary

The variation revealed in the distribution of variants of (r) in the sample demonstrates the complexity of the processes of linguistic change which the data suggest. Overall, although the non-localised [ j ] is found to be the most common variant for (r) in the data, an almost exact balance between the use of the localised variant, [r], and the spreading variant, [u], in both conversational style and word-list style is revealed. However, an analysis of the data broken down by age and gender reveals that the two distinct processes of accent levelling and diffusion can be seen to be acting simultaneously in the variation over apparent time revealed in one linguistic variable. The marked decrease in the use of the relatively localised [r] in the data can be ascribed to the process of levelling. The sudden emergence of [u] in the data from the young speakers of the sample may be explained by the process of diffusion. Further, the variation found in (r) reveals not only that two processes of linguistic change can act simultaneously in one variable, but it also suggests a gender difference between the initiators of these processes. The findings indicate that females lead in the levelling of variants, with males following, whilst males lead in the diffusion of new variants 173

into a variety, with females following. Moreover, in use of the non-localised [j ], we find the U-curve effect. If we recall the three patterns of distribution of variants across age outlined in section 3.1.3, we find that all three patterns of distribution are, in fact, in evidence in one variable.

In terms of the overall distribution of variants of (r) in MbE, it seems that changes are in progress and the situation in which a numerically equal distribution of the variants [r] and [u] exists is not a situation which will persist.

The data suggest that the localised variant, [r], is in the process of being levelled out, with similar findings reported in other localities of the North East of England, for example, Berwick upon Tweed (Watt & Ingham 2000) and Newcastle (Dominic Watt, personal communication). Additionally, if the trends revealed in the comparison of the adolescent data and the young adult data continue, that is if use of [u] continues to increase and use of [j] continues to decrease, [u] will become the preferred variant of (r) for young speakers, at the expense of [i]. The declining use of [r] and the increasing use of [u] across age suggest that the situation in which both variants are equally balanced in terms of levels of usage will not persist for too long, with the balance predicted to tip very soon to the side of [u].

Use of [u] appears not to be subject to style shifting and is variable for all speakers in the young groups, with no speakers demonstrating categorical use of the labio-dental variant. Additionally, a considerable amount of intra-group variability in the use of [u] exists in the four young speaker groups, with individual speaker scores ranging from 0% to 93.3%. Although the young males of the sample are ahead of the young females of the sample, use of [u] has clearly been adopted widely by the young speakers, with 14 of the 16 young speakers of the sample demonstrating some use of it. 174

4.4 (th) and (dh)

4.4.1 Variants of British English (th) and (dh)

The voiceless interdental fricative is found in only 18 of the 317 languages (6%) included in the UCLA Phonological Segment Inventory Database (UPSID) (Maddieson 1984). Ladefoged claims that ‘ [f] and [0] are pretty close acoustically and auditorally’, maintaining that ‘it is very rare to find them in a single language’ (in Fromkin 1985: 12). Indeed, of the 18 languages in which /0/ is found, only nine contrast /0/ and /f/ in native vocabulary (Maddieson 1984).

Although English is one of the nine languages to contrast /0/ and /f/, the replacement of the interdental fricative, [0], with the labiodental fricative, [f], is becoming increasingly common in BrE. The voiced interdental fricative 161 is more common in the world’s languages than is /0/, but it is still relatively rare

(found in 21 of the 317 languages in UPSID (Maddieson 1984)). English also contrasts 16/ and /v/, although again, despite the contrast, the replacement of the interdental fricative with the labiodental fricative is becoming increasingly widespread in current BrE.

As the interdental fricatives and labiodental fricatives are close acoustically and auditorally, they are relatively difficult to distinguish. Studies have found it to be harder to distinguish [Q] and [v] than it is to distinguish [0] and [f] (see further

Miller and Nicely 1955). Indeed, according to O’Connor (1973: 140), ‘even when perceptible friction is present for [v] and [6] it is the articulatory movements (or formant transitions) which provide the main cues for distinguishing between them’. The difficulty involved in distinguishing between the contrasting sounds combined with the fact that the interdental fricatives are rare in the world’s languages leads /0/ and /d/ to be considered ‘relatively unnatural segment types...learnt late by children’ (Wells 1982: 96). Wells goes further to state that ‘the labio-dental fricatives, [f] and [v], are more natural; and children, as is well known, readily substitute them for the difficult dentals’. The 175 continuing substitution of [f] and [v] for [0] and [6] respectively beyond childhood is thus described by Wells (1982: 96) as ‘persistent infantilism'.

The association of the use of the fronted variants, [f] and [v], with child speech, combined with the fact that the use of the labiodental fricatives for the interdental fricatives involves phonemic merger, has led to the overt stigmatisation of their use. This is reflected particularly in the media, agents of which constantly decry the use of TH-fronting as evidence of the deterioration of the language. In the popular view, this is often seen as a consequence of the seemingly unstoppable spread of what is termed ‘Estuary English’ throughout BrE. Interestingly, however, the use of TH-fronting has never been listed as a feature of ‘Estuary English’ (see further Rosewame 1984, 1994, Coggle 1993, Wells 1994). In a recent study which uses semantic differential scaling to test the social evaluation of TH-fronting (Costa de Oliveira Filho 1999), use of TH-fronting was found to be a feature associated with the South of England. (There is some suggestion that this may be changing, however, particularly amongst informants of Northern origin who perceived the feature as non-regional (Costa de Oliveira Filho 1999: 42).) The feature was also perceived as being found more commonly amongst under 30 year olds, as well as being a marker of male speech and an incorrect pronunciation generally associated with lower social classes and immature speech. In sum, it was ‘generally viewed as a stigmatised feature of southern lower class males’ (Costa de Oliveira Filho 1999: 72).

Apart from the stigmatised fronted variants of (th) and (dh), other dialectal or phonologically restricted variants are reported in urban varieties of BrE. Variants of the voiceless interdental fricative reported include [?0] in Newcastle (Watt and Milroy 1999), [?] in South East London (Tollfree 1999), [d} in Devon

(Hughes and Trudgill 1996). Scottish and Irish varieties of English appear to have considerably more variation, [h], [?], [s], [j ], [m], [n], [t] are all reported as variants of (0) (see further Stuart-Smith and Tweedie 2000, Chirrey 1999, Stuart -

Smith 1999, McCafferty 1999, Wells 1982). Variants of the voiced interdental fricative include [d] and [d] (particularly word-initially) for example in Derby

(Docherty and Foulkes 1999) and Sheffield (Stoddart, Upton and Widdowson 176

1999)). Again Scottish and Irish varieties of English appear to demonstrate more variation, with variants including [r], [d], [1], [0], [m], [h] (see further Stuart-

Smith and Tweedie 2000, Stuart-Smith 1999, McCafferty 1999, Wells 1982). Additionally, [0] and [d] are sometimes elided. The elision of intervocalic [6] is a feature of Northern Ireland English (see further Milroy 1987a, McCafferty 1999). Zero realisations are also noted in RP due particularly to the ‘difficulties o f articulation when followed by /s,z/’ (Cruttenden 1994: 168). In the present study, however, it is the phonemic mergers involved in the use of [f] for (th) and [v] for (dh) only which are of interest in the data from Middlesbrough.

4.4.2 Geographical and social distribution of variants under consideration

The replacement of the interdental fricatives by labiodental fricatives is generally associated with the speech of London or ‘broad Cockney’ (O’Connor 1973: 141, Cruttenden 1994: 168, Crystal 1995: 327, Hughes and Trudgill 1996: 70, Wells 1982: 328). Indeed, such is the association of TH-fronting with Cockney or London English, that it is used to dramatic effect in the title of Barltrop and Wolveridge’s 1980 work on Cockney life and language, The Muvver Tongue.

Although written evidence exists which suggests that TH-fronting could be found in London English in the first half of the nineteenth century (see Machyn 1843: 134), the spread of TH-fronting from London to urban centres further removed from the capital is considered a relatively new phenomenon which has increased dramatically in recent years. Evidence from the SED, however, suggests that the fronted variants of (th) and (dh) could be found in the Midlands and even the Northern Counties in addition to the Southern Counties in the middle of the twentieth century. Although use of the fronted variants of (th) and (dh) was considerably more widespread in the Southern Counties and the East Midlands, the pronunciation [Jiaf] for sheath is reported for localities in Yorkshire1. Indeed, use of [f] for (th) is reported as a feature of West Yorkshire English from a much

1 The Yorkshire localities of the SED in which TH-fronting was recorded do not include the two which are nearest to Middlesbrough, Stokesley (Y2) and Skelton (Y3). 177 earlier date than the SED data. Joseph Wright, in his 1892 work on Windhill in the West Riding of Yorkshire, reports that use of TH-fronting in the first half of the nineteenth century was, in fact, more prevalent than in the 1890s.

Fifty years ago f for J) [=0] and v for d were quite general throughout the Township of Idle, but they have now practically disappeared except as an individualism (Wright 1892: 91).

The use of TH-fronting as a consequence of the perceptual confusion of the sounds is also noted by Henry Sweet (1874), as reported in Ohala (1993: 242): ‘Sweet (1874: 8), among others, noted the frequent confusion and consequent substitution of [f] for [0] in dialectal English, e.g. [0iq] ~ [fig] 'thing'; [0ju] ~

[fiu], 'through'.

Given the age of the informants used in the SED, coupled with reports of the frequent incidence of TH-fronting from Joseph Wright and Henry Sweet, it may be that TH-fronting was considerably more widespread in early nineteenth century English, particularly in Northern England, than was the case in twentieth century BrE. It may also be the case that, given the ‘general’ and ‘frequent’ incidence of TH-fronting in the nineteenth century as evidenced by Wright and Sweet, the feature may not have suffered the overt stigmatisation in the past that it has in recent years. Nonetheless, in recent studies of urban varieties of BrE there seems little doubt that TH-fronting is a feature which is associated with and indeed used by young speakers. As such it is usually argued that TH-fronting is a feature which is new to the variety under investigation and, further, it is generally a feature which receives overt stigmatisation. Such is the speed and breadth of the diffusion of TH-fronting to urban centres of Britain in recent years that Trudgill (1999a: 137) claims ‘TH-fronting in modem English is a remarkable phenomenon’, and that ‘formerly confined to the London area and to Bristol, it has in the 1980s and 1990s begun to spread enormously rapidly across England’.

Although TH-fronting was reported as being heard in Leeds by Wakelin (1972) and Sheffield by L. and J. Milroy in the 1980s (Milroy 1996), it is still generally associated with London and the South-East of England, as noted. The rapidity of 178 its diffusion into a variety and its association with young speakers is demonstrated by the fact that TH-fronting was found to be near categorical in the speech of some children in Norwich from the 1983 survey (Trudgill 1988: 43). Although it had been totally absent from the 1968 fieldwork as reported in Trudgill (1974), with none of the informants born before 1958 showing any TH- fronting, 70% of informants born between 1959 and 1973 had some degree of TH-fronting, with 29% revealing no instances of [0] at all (Trudgill 1999a: 138).

Kerswill and Williams (1997: 189) found that children in Milton Keynes had a 68.3% use of [f], with all young speakers using at least some [f], whilst a 17% use only was revealed by their caregivers. 12 year old children were reported as having an 84% use of [v], with caregivers showing a 4% usage (Kerswill & Williams 1997: 238). Also, in Sandwell in the Midlands, TH-fronting is found for an ‘increasing number of teenagers’ and, indeed, is ‘nearly categorical with some boys’ (Mathisen 1999: 111)

Considering more northerly locations, TH-fronting was evident in the recent study of Derby English (Milroy 1996). Findings revealed a marked difference between generations, with TH-fronting found in the speech of the young informants, but not in the speech of the older speakers. Further, a clear social class difference was revealed, with TH-fronting appearing in the speech of the young WC informants (62%) to a considerably greater extent than in the MC young speaker group (7%). Although the male speakers favour the [fj and [v] variants more than do the females, the difference found was slight (Milroy 1996: 216). In Hull, use of the fronted variants of (th) and (dh) was shown to be near categorical in the working class male group of adolescent speakers (90.7% use of [f], 95.5% use of [v] (Williams & Kerswill 1999: 160)). Indeed, the level of use of TH-fronting for this particular speaker group was higher in the northern city of Hull than in the two southern locations of the study (Milton Keynes and Reading). WC females in Hull revealed a lower score than their male counterparts (63.2% use of [f], 77.7% use of [v]). Use in the MC group was considerably lower still. Similarly, [fj and [v] are reported as possible realisations of (th) and (dh) for ‘younger speakers, particularly males’ in Sheffield (Stoddart, Upton and Widdowson 1999: 76). In Newcastle, also, the 179 labiodental forms can be found, although they are still ‘relatively scarce’ (Watt and Milroy 1999: 30). Additionally, in Glasgow [f] and [v] are found ‘variably but frequently in the speech of WC children, in both word-list and conversational speech’ (Stuart-Smith 1999: 209).

4.4.3 Particulars of the data and areas of interest

In the conversational Middlesbrough data, 30 tokens were sought for all speakers for both variables (th) and (dh). Word-initial, word-medial and word-final positions were analysed for (th), for example, think, nothing, both. However, word-medial and word-final positions only were analysed for (dh), for example, bothered, seethe, as, according to Wells (1982:328) and Milroy (1996: 215), TH- fronting does not occur in word-initial tokens of (dh). This was borne out in the word-list data in the present study. Three tokens of word-initial (dh) were included in citation form, with the result that no word-initial tokens were produced with a [v] realisation. Analysis of word-medial and word-final tokens of (dh) only resulted in the availability of fewer tokens per speaker. Therefore, it was not always possible to achieve the desired quota of 30 tokens per speaker for (dh).

In the conversational data from Middlesbrough, variants of (th) which were recorded were: [0], [f], [3], [?], [?0], [n], [t]2 and a zero realisation. Variants of

(dh) in the conversational data analysed were: [3], [v], [0], [d] and a zero realisation. All realisations apart from [0] and [f] for the voiceless fricative and

[3] and [v] for the voiced fricative were infrequent, seemed to be lexically conditioned, and appeared not to show any sociolinguistic patterning. As TH- fronting is the feature of interest to this research, this section discusses use of [0] and [f] realisations of (th) only and [3] and [v] realisations of (dh) only. The

2 Interestingly, although [t] is mentioned as a feature of the Cleveland dialect by Atkinson (1869: xiv) and recorded as a regional feature by Upton & Widdowson (1996) based on the findings of the SED, only one [t] realisation of (9) was recorded from all speakers in the sample. 180 scores presented in this chapter, therefore, involve an analysis of 30 tokens from each speaker in which [0] and [f] for (th) and [6] and [v] for (dh) only were included. An analysis of 30 tokens from each speaker, in which all realisations of (th) and (dh) were included, was also undertaken, however. The extent of the use of the other variants recorded in the conversational data across speaker groups is presented as Appendix 9. The marginal use of the other possible realisations of (th) and (dh) is demonstrated by the fact that for all realisations (apart from [0] and [f] for (th) or [8] and [v] for (dh)) scores of under 10 tokens of 120 from each speaker group are recorded. Indeed, in most cases the score is under 5.3

In citation form, six instances of (th) were included. These involved (th) in word- initial, word-medial and word-final positions. Apart from the three word-initial instances of (dh) (which are omitted from the word-list data to follow), six instances were included in citation form. These included (dh) in word-medial and word-final positions.

The focus of interest is whether the fronted variants of (th) and (dh) are present in the data from Middlesbrough and, if so, which speakers reveal use of them. Given the scarcity of their use in Newcastle, and the fact that speakers from urban centres closer to London appear to show a considerably higher use of [f] and [v] than those in urban centres further removed from London, a comparable amount of TH-fronting to that found in Tyneside English may be expected.

Before analysing the data in terms of correlations with particular speaker groups, the overall distribution of the labiodental variants in the data is considered.

3 The zero realisation of (th) was the most frequent variant other than [0, f] and [5, v]. This was largely found in the word clothes, however, in which elision is likely, as noted (see Cruttenden 1994: 168). 181

4.4.4 Overall distribution of variants of (th) and (dh)

Table 4.16 below presents the overall use of the interdental and labiodental variants of (th) and (dh) in the conversational data.

Table 4.16: Overall distribution of variants of (th) and (dh) (conversational style, all speakers)

Total [9] [5] W [v] n. % n. % 1769 1590 89.9 179 10.1

The interdental variants are revealed to be the preferred variants by far across speaker groups, accounting for almost 90% of the conversational data. Although they account for just 10% of the data, the fronted variants are in evidence in MbE, however. Whether their use is associated with any particular speaker group will be shown as we turn to an analysis of the (th) data broken down by age, then age and gender in conversational style. We then examine the distribution of variants of (dh) in the same manner, before analysing the word-list data.

4.4.5 (th), conversational style

4.4.5.1 Generational variation

Table 4.17: Distribution of variants of (th) across age

[0] m n. % n. % Old 240 100 0 0 Middle 240 100 0 0 Young 377 78.5 103 21.5 182

[f] is revealed to be in evidence in the speech of the young only, suggesting that [f] is a new variant which is being introduced into MbE by the young speakers. No instances of [f] are noted in the 480 tokens recorded from the old and middle groups, emphasising the sudden emergence of the new variant into the speech of the young. Furthermore, a substantial amount of [f] usage is noted. We turn to an analysis of the distribution of variants of (th) amongst young speakers only, to observe whether gender-correlated variation is in evidence in the data and whether fine-grained age differentiations are observable between the adolescent and the young adult speakers.

4.4.5.2 Age and gender

Figure 4.25 below presents the distribution of variants of (th) in the speech of the 16 young speakers broken down by age and gender.

Figure 4.25: (th), age and gender (young speakers)

98.3 97.5 97.5

young adolescent young adolescent adult female (N. adult male male (N. female (N. 120) (N. 120) 120) 120)

Figure 4.25 clearly shows that one group of speakers is virtually wholly responsible for the use of [f] found in the conversational data. That group is the young adult males. All other groups of young speakers reveal some use of the fronted variant, but use of [f] amongst the young adult females, the adolescent females and the adolescent males is marginal. The level of use of [f] amongst the young adult males in comparison is remarkable, and at almost 80% it is by far the 183

preferred variant of the young adult male speakers. All four young adult males reveal [f] to be their preferred variant. Whether the same patterning is revealed in use of (dh) is considered below.

4.4.6 (dh), conversational style

4.4.6.1 Generational variation

Table 4.18: Distribution of variants of (dh) across age

[5] [v] n. % n. % Old 240 100 0 0 Middle 218 100 0 0 Young 275 78.3 76 21.7

The pattern found in the distribution of variants of (th) is also in evidence in the data for (dh). [v] is also a new variant in the data which is only in evidence in the speech of the young in the sample. Again, [v] is completely absent from the speech of the old and middle groups. Furthermore, the level of use of [v] amongst the young is as substantial as use of [f], with virtually the same level of use recorded (21.5% use of [f] and 21.7% use of [v]). As with the results for (th), examination of the distribution of variants of (dh) amongst the four young speaker groups only will now be undertaken to ascertain whether the same unusual trend is in evidence in the distribution of variants of (dh) as was found in for (th).

4.4.6.2 Age and gender

Figure 4.26 presents the distribution of variants of (dh) amongst the 16 young speakers broken down by age and gender. 184

Figure 4.26: (dh), age and gender (young speakers)

n interdental ■ lablo-dental

young adolescent young adolescent adult female (N. adult male male (N. female (N 74) (N. 96) 94) 87)

The same patterning is indeed revealed in (dh) as that found in (th). Again, one group of speakers is seen to be overwhelmingly responsible for the use of the fronted variant in the data, that group being the young adult males. All groups of young speakers reveal some use of the fronted variant of (dh). However, again, the young adult female, the adolescent female and the adolescent male use of [v] is marginal. Although not quite as high as their use of [f], the young adult males’ use of [v] for (dh) is also remarkable in comparison with the level of use of the variant amongst all other speakers of the sample.

4.4.7 (th) and (dh): citation forms

Six tokens of (th) and six tokens of (dh) were included in citation form to ascertain whether TH-fronting was subject to style shifting. Table 4.19 below presents the overall distribution of variants of (th) in the word-list reading.

Tabic 4.19: Overall distribution of variants of (th) and (dh) (word-list style)

Total [0] [6] m M n. % n. % 370 352 95.1 18 4.9 185

Use of [0] and [5] is virtually categorical in word-list style. This was the pattern

found in the conversational data. Use of [0] and [9] is higher in word-list style,

however. Consequently, use of [f] and [v] is lower. It should, however, be noted that within the word list data five ‘in between’ realisations were noted which were not included in the scores presented4. The fact that labiodental realisations are in evidence in the word-list data would suggest that the speakers who produce [f] and [v] in conversational style are also the speakers who produce [I] and [v] in word-list reading. We turn then to an analysis of the use of [f] and [v] in the word-list data, broken down into the eight speaker groups to ascertain whether this is indeed the case.

Table 4.20: Distribution of [f] and [v] (word-list style (WLS) and conversational style (CS))

m % [v] % WLS CS WLSCS Om 0 0 0 0 Mm 0 0 0 0 YAm 35 79.2 50 71.9 Adm 0 2.5 0 3.2

Of 0 0 0 0 M f 0 0 0 0 YAf 0 1.7 0 1.1 Adf 0 2.5 0 4.1

All speaker groups apart from one avoid the fronted variants categorically in word-list style. The one group revealing some use of [f] and [v] is the young adult male group; the group of speakers revealed to be virtually wholly responsible for use of the fronted variants in the conversational data. The

4 Realisations of (th) and (dh) which were felt to be between an interdental articulation and a labiodental articulation were also recorded in Derby (Milroy 1996) and in Glasgow (Stuart-Smith and Tweedie 2000). 186

marginal use of [f] and [v] in the other young speaker groups is not in evidence in word-list style. Incidence of [f] and [v] is lower in word-list style as compared with conversational style amongst the young adult males, suggesting a degree of style shifting. Given the overt stigmatisation carried by the fronted variants, however, the degree to which they are in evidence in the formal speech style is still quite surprising. All four young adult male speakers make some use of the fronted variants in word-list style.5 This is further evidence which suggests that the young adult males conform less to the usual shift to a more formal speech style in word-list reading.

4.4.8. (th) and (dh): summary

Unusual results have been revealed in the data presented for (th) and (dh). Evidence suggests that [f] and [v] are new forms which have emerged very recently in MbE. The speakers of the sample found to be responsible for their use are therefore of particular interest. Consequently, to find such a specific group of speakers, the young adult males, almost wholly responsible for use of the new forms in the data raises issues of interpretation concerning the profile of speakers responsible for the diffusion of new variants into a variety and the motivation for the adoption of overtly stigmatised variants, which will be considered further in Chapter 6.

No difference in the level of use between the voiceless and the voiced fronted variants in the conversational data is observed (21.5% use of [fj, 21.7% use of [v]). Also, little intra-group variability is in evidence in the findings. All of the speakers from the young adult male group reveal a substantial level of use of [ f] and [v] in conversational style. Likewise all of the speakers of all other groups show categorical or near categorical use of [0] and [&]. There is some evidence

of style shifting in the data, although substantial use of [f] and [v] is also revealed by the young adult males in the more formal word-list style. This may be due

5 Interestingly, all four members of the young adult male group produced an [f] realisation of (th) in the citation form everything. This was the only citation form in which all speakers produced the fronted variant. 187 largely to this particular group of speakers’ lack of style shifting in general, which we shall also consider further in Chapter 6. 188

4.5 Summary of trends

The results presented in this chapter are illustrative of several trends in MbE. The distribution of variants of all variables analysed has been revealed systematically to correlate with the social characteristics of the speakers of the sample. In all variables under investigation, age- and gender-correlated variation is observed. Details of which forms are being introduced into MbE as well as those which appear to be increasing or decreasing from old to young are outlined below.

Variants which can be seen as innovatory forms; they appear exclusively, or near exclusively in the speech of the young in the sample:

• [u] as a realisation of (r)

• [f] as a realisation of (th)

• [v] as a realisation of (dh) • [ht] as a realisation of pre-pausal and turn-final (t)

Variants which can be seen to be increasing; a higher level of use is revealed in the speech of the young as compared with the middle and older speakers: • [?] as a realisation of word-medial (t)

• [?] as a realisation of pre-pausal and turn-final (t)

• [?p] as a realisation of word-medial (p)

Variants which can be seen to be decreasing or being lost; a decreasing level of use is observed in the speech of the young as compared with the speech of the middle and old speakers: • [r] as a realisation of (r) • [t] as a realisation of word-medial (t) • [t] as a realisation of pre-pausal and turn-final (t)

Three of the innovatory variants in the data are forms which constitute part of the current vernacular changes in BrE. One innovatory form, however, does not. Pre­ aspirated pre-pausal and turn-final (t) can be seen as a North Eastern innovation 189 as its use is thus far recorded in Newcastle English only. In terms of localised forms being lost, thus leading to homogenisation of BrE, we find only one form which can be modelled in this way; [r] for (r). Use of other localised forms is increasing in the speech of the young. The other forms which are decreasing and being lost are in fact the prestige or standard forms, whilst non-standard, stigmatised forms are increasing.

Evidence suggestive of change in progress is found in almost all linguistic variables of the study. The U-curve effect, the expected distribution of variants observed when a variable is not undergoing change, is found in (k) and (p). However, interestingly, the gender-correlated variation in evidence in the variable (p) suggests that a variable can be both stable and undergoing change, (p) is a stable variable in the speech of the males of the sample, where, in terms of use of the localised variants, we find a U-curve distribution of variants. However, it is in the process of change in the speech of the females, where we find a sharp increase in the use of the localised variant. Although we find evidence suggestive of convergence with varieties further north in (p), it seems that when a localised, or North Eastern, variant is in competition with a form which is spreading nationally, one seen arguably as a youth norm, the youth norm will win out. (We do not see a corresponding increase in the use of the localised [?t] for word-medial (t) where [?] is used near categorically by the young of the sample.) In terms of processes of linguistic change, then, apart form the increased use of the localised variant of (p), which suggests convergence with varieties further north, the processes of levelling and diffusion are also in evidence. Indeed, the variation evident in (r) indicates that these two processes can act simultaneously in one variable. The decreasing use of [r] in the speech of the young of the sample is attributable to the process of levelling, whereas the adoption of [u], also in the speech of the young, can be ascribable to the process of diffusion.

Gender-correlated variation is found for all variables. In terms of gender and processes of linguistic change, from the results presented in this chapter it can be argued that female speakers lead in the process of levelling (note the lower 190

female level of use of [r]) and also in the increased use or adoption of forms with

supra-local currency (note the higher increase in female level of use of [?p] and

adoption of [ht]). Males, on the other hand, lead in the adoption of forms which

are spreading nationally (note the higher male use of [f|, [v] and [u]).

Apart from the generational differences uncovered in the variables which are suggestive of linguistic change in progress, fine-grained age differentiations have been uncovered in the data from the young adult and adolescent speaker groups. The fine-grained age differentiation in evidence reveals that different speaker groups from the young speakers of the sample are responsible for the adoption of innovatory forms. The two groups found to innovate are the adolescent females and the young adult males. Finally, in terms of style shifting, an important finding appears to be that the speakers who innovate are less likely to style shift.

The results presented in this chapter will be considered further in Chapter 6, where motivation for the linguistic variation observed will be considered and interpretation of the trends uncovered will be offered. Such interpretation will be arrived at through insight gained from attitudinal information which is the subject of Chapter 5. Chapter Five Results: Perceptions of Language and Place

5.0 Introduction

Attitudinal data and speakers’ perceptions of language and place are seldom incorporated into quantitative studies of language variation. However, in order to gain insight into the motivation for linguistic variation and for the change in progress it can imply, attempts must be made to access the local knowledge that speakers operate with when constructing and projecting their sociolinguistic identities. To this end the attitudinal data procured through the Identity Questionnaire (IdQ) section of the interview are presented in this chapter. As well as providing insight into the construction of community and the collective identity of the informants, the responses to the questions posed in the IdQ reveal evidence of age- and gender-correlated variation within this collective identity construction. Therefore the attitudinal data work on two levels. The first level is concerned with what can be said to constitute community identity in Middlesbrough. The second level considers cohort identity construction which is perceptible or above the level of awareness within this community identity.

The chapter begins with an analysis of informants’ perceptions of MbE. How MbE is defined and delimited is analysed, as are perceptions of variation within MbE. Perceptions of place are then presented, with attitudes towards Middlesbrough explored. The strength of local affiliation felt by informants is also examined through assessment of the informants’ overall reactions to the IdQ section of the interview and their rating on the Affiliation Score Index as outlined in section 3.2.3.3. Interpretation of the attitudinal data and how they correlate with the linguistic data presented in Chapter 4 will then be considered in Chapter 192

5.1 Language

5.1.1 Defining and delimiting MbE

Insight into how the informants of the study define and delimit MbE is provided by the responses to the following questions of the IdQ: • What accent would you say you had, and do you like it? • Can you recognise the accent of Middlesbrough (e.g. if heard on the radio or TV)? If so, how? • Where, geographically, would you say people stop talking the same as you and start sounding different?

5.1.1.1 Defining the accent

The first question seeks to establish what label the informants would give themselves. Responses to this seemingly straightforward question prove telling. Variation in the responses to the first question is shown in all groups of speakers. However, the majority response of each age group tallies exactly with the history of the shifting identity of Middlesbrough, as outlined in section 3.1.1.4 (see Table 5.1 below).

Table 5.1: Definition of accent across age Yorkshire Teesside M’bro North/NE total n. % n. % n. % n. % o 8 4 50 1 12.5 3 37.5 0 0 M 8 0 0 4 50 3 37.5 1 12.5 Y 16 0 0 3 18.75 11 68.75 2 12.5

Amongst the older speakers, the most frequently given response is ‘a Yorkshire accent’. Most of the older speakers who defined their accent as Yorkshire qualified their response by stressing that it was not a ‘broad Yorkshire’ accent. This, combined with responses such as ‘I would call it Yorkshire because I was born in Yorkshire’, suggest that accent can be defined by geographical place 193 regardless of whether it conforms to the speaker’s perception of the accent in question. The most frequently given response of the middle aged speakers is 'a Teesside accent’ further emphasising the importance of geographical place in terms of definition of accent. Amongst the combined group of young speakers the most frequently given response is ‘a Middlesbrough accent’, it being given by 11 of the 16 young speakers. This suggests that speakers react to the changing political boundaries of the area in which they live. If the political boundaries of the area change, so will how inhabitants perceive themselves. The identity of the speech community can be seen to be realised through the individual inhabitants’ perceptions of what accent they have and where they come from. Circumstances can be such as to change the labels used to describe an area and its people. This can change the labels inhabitants use to describe themselves, which can, in turn, change the label attached to the speech community.

5.1.1.2 Identifying the accent

The second question of this section seeks to establish whether speakers of MbE can identify salient features of MbE which make it recognisable. Unsurprisingly, this proved a difficult question on which to elaborate. Nonetheless, responses reveal interesting insights both in terms of age- and gender-correlated differences in the perceived ability to recognise MbE, and in terms of the features which are considered indicative of the variety in question.

Of the eight older speakers, seven claimed the ability to recognise MbE, with all of the females answering in the affirmative. Similarly, amongst the speakers of the middle group seven of the eight claimed to be able to recognise their accent. However, in the middle age group the female speakers were considerably more tentative in their replies, and qualified their answers with responses such as:

‘you can tell it’s northern, but it’s not Geordie by any stretch is it? - so it’s sort of it’s northern but it’s not Yorkshire - so it’s sort of in the middle’.

In the combined group of young speakers only 10 of the 16 informants answered in the affirmative, with six claiming that they could not recognise a 194

Middlesbrough accent. This is interesting in view of the responses to the previous question of this section, in which 11 of the 16 young speakers claimed that they would define their accent as a ‘Middlesbrough accent’. Five of the six young speakers claiming not to be able to recognise a Middlesbrough accent were female. Furthermore, two of the female speakers who responded in the negative alleged that the similarity of the accent to Geordie was the reason they would not be able to recognise MbE. Overall, then, the male speakers appeared more confident than the females in their ability to recognise MbE if heard through the media. The perceived inability amongst female speakers, particularly young female speakers (three of the four adolescents and two of the four young adults responded negatively to the question), to recognise the accent may be linked to the suggested female role in accent levelling and the development of supra local norms (see further Watt & Milroy 1999). This will be considered further in Chapter 6.

From the responses to the second question, other than words and expressions, the most frequently cited features seen as central to the recognition of MbE are vowels. This is commented on in a general sense as in the mention of ‘flat vowels’ by a number of informants, and in comments such as ‘it’s got its own sound, hasn’t it? - it’s definitely got its own vowel sounds or something’. However, more specific features are also referred to. An [e:] pronunciation of the n u r s e vowel is mentioned by a number of informants as indicating a Middlesbrough accent. This is considered a feature of Liverpool English, as noted in section 3.1.1.2. Consequently, use of this variant may contribute to the similarity of MbE and Liverpool English as evidenced in the recent study by Kerswill and Williams (2000) in which the Middlesbrough accent was mistakenly identified as a Liverpool accent by a number of informants, as noted in Chapter 2. Indeed, some speakers in the present study recounted the experience of being mistaken for Scouse, the accent of Liverpool, and awareness of the similarity between Liverpool English and MbE is in evidence in the data with responses to the second question such as ‘it’s halfway between Scouse and

Geordie’. The monophthongal realisation of the f a c e vowel, [e:], is mentioned as

being indicative of MbE as opposed to the centring diphthong [ia] which is 195

perceived as denoting a Geordie accent. The GOAT vowel is also mentioned and described as being ‘long’, presumably indicating a monophthongal realisation. Interestingly, a [bu:k] pronunciation of book is also cited as indicative of MbE as

opposed to the [buk] pronunciation said to be found in nearby Redcar. Other than vowel sounds, the feature most frequently cited as denoting MbE is the use of clause final like}

5.1.1.3 Delimiting the accent

The related question of delimiting MbE was put to the informants in the third question under consideration in this section. Figure 5.1 below illustrates all locations mentioned by informants in response to the question of where, geographically, people are perceived to have a different accent to that of the informants.

1 Evidence of the salience of clause final like in MbE can be found in its use in the title o f an article about an accent-related story in the Middlesbrough Evening Gazette. The article was entitled ‘Knoworrameanlike!’ and in the accompanying ‘essential guide to Tees Speak’ like was listed with the instruction ‘slap it on the end of any sentence’ (Evening Gazette, April 23, 1999). 196

The most frequently mentioned location in response to the third question is Hartlepool, which is cited 10 times, followed by Sunderland which is referred to nine times by informants. As can be seen in Figure 5.1, the perceived delimitation boundary appears closer south of Middlesbrough than it is north of the town, suggesting that informants perceive MbE as more similar, or less different, to the varieties found in the North East than those found in North Yorkshire.

The perceived definition and delimitation of MbE, then, is a complex matter. Any definition of Self immediately evokes Other. The construction of a collective identity in terms of defining and delimiting the variety of BrE spoken in the community under investigation is undertaken through the contrasting of Self (that is, members of the speech community) with Other (members of other speech communities). In this way, the three questions in this section relate directly to the first level of identity construction, that at the level of community. Additionally, however, the responses to the questions also relate to the second level of identity construction, that of discernible cohort identities within the community identity, as age-correlated variation is in evidence in the definition of accent. The implications of the differing definitions of Self, as well as the grouping of Self and the delineation boundary of Other, will be considered further in Chapter 6.

5.1.2 Perceptions of variation in MbE

Insight into perceptions and awareness of variation within MbE is sought in the responses to the following questions: • Do you think older and younger people talk the same here (pronounce things the same and use the same words)? • Do you think there’s a difference between how males and females speak here? 197

5.1.2.1 Perceived age variation

The first question of this section seeks to ascertain speakers’ awareness of age- correlated variation which, in turn, indicates the extent of speakers’ awareness of linguistic change in progress in MbE. Responses prove illuminating both in terms of differences which are noticed in speech and those which are not. Of the 32 informants in the sample, 28 responded to the question with the claim that they are aware of age-correlated variation in MbE. The majority of speakers, however, elaborate on this affirmative response with reference to lexical differences. Lexical variation is not dealt with in the present study. Therefore, discussion of responses to this question will be confined to speakers’ perceptions of variation in phonological features.

The perception that young speakers ‘miss letters out’ is commented on by 10 speakers of the sample. In elaboration of this point, the ‘missing out of letters’ is generally found to refer to the use of [?] for intervocalic /t/, suggesting that use of the glottal stop realisation of /t/ is well above the level of awareness. This will be discussed in more detail in the following chapter. Five of the 16 young informants comment on the ‘dropping of letters’ by young speakers. Only one of the middle speakers mention the ‘dropping of letters’ in the speech of the young, whilst it is mentioned by four of the eight older speakers (three females). Other than use of the glottal stop realisation of /t/, evidence of the awareness of two other innovations is revealed in the data, [f] and [v] realisations of /0/ and 161 respectively are mentioned by two speakers, both older females. Furthermore, use of the fronted variants of /0/ and 16/ is considered by these informants to be incorrect and would be subject to correction when heard. Additionally, the use of intrusive /r/, which is claimed to be a ‘southern thing’, is referred to by one older female speaker. Interestingly, however, use of a labiodental approximant for /r/ is not referred to. Implications of these responses will also be discussed in the following chapter. In terms of the loss of features, as evidenced in the awareness of forms used by older speakers but not by younger speakers, monophthongisation is mentioned, with one young speaker offering the example of [juaz] as used by older speakers compared with [jd:z] as used by younger 198 speakers. Similarly, a [bu:k] pronunciation as used by older speakers as opposed to [buk] by younger speakers is referred to by a young informant. Age-correlated variation in speech rate and intonation is also mentioned by some informants, with young speakers perceived as speaking at a faster rate than older informants.

In general terms, the opinion that older speakers are more precise and well- spoken than younger speakers is expressed by a number of young informants. This is seen by some as being the result of a conscious effort, with opinions like ‘probably when they get older, they think they have to talk a bit more posh’, and old speakers are ‘putting it on to sound posh’. Conversely, younger speakers are considered ‘more common’ than older speakers by a number of informants, and their language is considered to be ‘harder’. Some informants stress the influence of television on the speech of the young. Insightful comments are also made on the importance of accommodation in age-correlated variation, with reasons for young speakers’ use of low prestige or non-standard variants being attributed to the fact that ‘they don’t like to be different - it’s easier to fit in than be different’. Similarly, awareness of the decreasing use of low prestige or localised variants in the middle, work-oriented years, which corresponds to the U-curve effect (McMahon 1994: 241), is attributed to the fact that:

‘when you go to work, you realise that you have to be understood - you mix with different people, not just your peer group’.

Another observation of particular relevance is that of two of the young adult females of the sample whose immediate response to the question was to stress the perceived difference between themselves and adolescent speakers, with one speaker stating that ‘younger people are even more common than we are - even I have trouble understanding them sometimes’. Thus, the speech of teenagers is considered by these two 21 year old females to be more localised or more non­ standard than the speech of young adults. This observation will be explored further in Chapter 6. 199

5.1.2.2 Perceived gender variation

The second question of this section seeks to gain insight into speakers’ awareness of gender-correlated variation in MbE. Responses are illuminating, and interesting variation across age groups towards perceived gender differences in speech is revealed. This is particularly enlightening in terms of attitudes shown towards the speech of females. Amongst older speakers, those who feel that a discernible gender difference in speech can be observed express this difference in terms of males being ‘broader’ or using a more localised accent than females. An example can be seen in the comment of an older female:

‘probably men are a bit broader I should think - women try to make themselves a bit refined sometimes don’t they, more than men - it’s part of kind of an image thing maybe’.

However, this female concern with image in terms of speaking ‘correctly’ is not seen as extending into the speech of young females, as it is felt, by the same speaker, that ‘girls are more aggressive now’.

In contrast to the older speakers, none of the speakers from the middle group claim to be aware of gender differences in speech. However, the reasons given for this apparent similarity are enlightening. All four males from the middle group claim that there are no differences between males and females because the females are as broad as males. Although in all four middle males’ responses it is claimed that no gender difference can be perceived, in their responses it is clear that a difference is felt. An example can be seen in the following response to the question, ‘not really no, when you listen to some women sometimes they’re more rough and more broad than some of the blokes, aren’t they?’ [my emphasis]. The speaker thus describes a difference immediately after denying one exists. This suggests that the only conceivable difference or the only difference that would be acknowledged is a difference of males making greater use of localised or low prestige variants than females. Furthermore, amongst the males of the middle group the perception of female speakers using as localised an Accent as males is evaluated negatively. Examples include: 200

‘in ‘basic’ areas of the town you hear the girls talking just like the lads — it sounds crude - it sounds wrong’

‘I don’t think there’s any [difference] when I listen to some of the fishwives round here’.

Interestingly, none of the female speakers from the middle group claimed to perceive a gender difference.

Amongst the young speakers, the picture is different again. More speakers from the young groups claim to perceive a discernible gender difference in MbE than is found amongst the middle or older speakers. Perhaps the most interesting group in this respect is the young adult males. Three of the four members of this group maintain that a considerable difference is perceivable. This difference is felt to lie in the perception that female speakers are notably broader and ‘more common’ than males. However, this only applies to young females, to the extent that two of the young adult males specify an age range within which this difference is discernible (between the ages of 12 and 18). It is felt that amongst older speakers no gender difference is observable except in terms of willingness to accommodate towards a less localised style of speech, as:

‘a woman’d probably try and change it more than a bloke - a bloke probably wouldn’t be as bothered whereas a woman would make an effort’.

Again, this perceived female use of low prestige variants is evaluated negatively by the young adult males, with opinions such as ‘females all need their mouths washing out with soap and water round here’. Surprisingly, amongst the adolescent males, opinions are not so definite, although, again, the opinion of young females being ‘common’ is expressed. Amongst the female speakers of the young group, differences are expressed in terms of the use of different phrases. Only one of the eight young females claimed that males are ‘more harsh’ than females, maintaining that male speakers ‘miss letters off words’ more than females. None of the young females expressed the idea that female speakers are broader than males, however. 201

From the responses to this question we can see the progressive changing of opinions towards female speech. The older speakers concur with the traditional view that female speakers deliberately use a greater proportion of high prestige variants than males. However, across age a change in the perception of how closely the female speakers conform to this image is observed, particularly amongst the male speakers. The males from the middle group no longer detect a difference between the genders, and the direction of the perceived change is what is of interest. It is not that the male speakers are perceived as using a greater proportion of high prestige variants, as perhaps would be expected from their work-oriented life-stage and their supposed position in the U-curve as noted in Chapter 2. Rather, the change in opinion is felt to be due to the female use of a comparable proportion of low prestige or localised variants to that of the males themselves. This opinion could be accounted for with recourse to the fact that speakers from the middle group are thought to be overly concerned with notions of correctness. Alternatively, it could be because the speech of females is changing, with female speakers increasingly using a higher proportion of low prestige variants. Thus the perception of gender differences has changed over time. This argument is strengthened by the fact that the young males of the sample not only consider females to be as broad as males, but in fact considerably broader. These changing perceptions of female speech will be explored further in the following chapter.

Thus, we have evidence that within the community identity which is shared by the informants, there is awareness of cohort identities which manifest themselves in linguistic behaviour. We not only see perceptions of variation within MbE, but we also find evidence of the evaluation of this variation. Furthermore, differences in these perceptions and evaluations are in evidence, and these differences also appear to correlate with the age and gender of the informant.

5.1.3 Attitudes towards MbE

Insight into the informants’ attitudes towards MbE is sought in the responses to the following questions: • What accent would you say you had, and do you like it? 202

• Have you ever been in a situation where you’ve deliberately changed the way you talk? If so, why? • What would you think if your accent was referred to as Geordie or Yorkshire?

5.1.3.1 Evaluation of the accent

The question of whether informants like their accent is posed in the second part of the first question of this section. Overall, almost half of the sample responded in the affirmative, with 15 of the 32 informants stating that they like their accent.

Table 5.2: Informants’ evaluation of accent (male/female) Affirmative Non-committal Indifferent Negative Total n. % n. % n. % n. % Male 16 10 62.5 1 6.25 3 18.75 2 12.5 Female 16 5 31.25 2 12.5 3 18.75 6 37.5

As regards gender variation, a higher proportion of male speakers than female speakers claim to like their accent, with 10 of the 16 males responding in the affirmative compared with five of the 16 females, as can be seen in Table 5.2. At times the positive response is qualified by comparing MbE favourably with a Geordie accent, which is felt to be incomprehensible, for example, ‘we’re big­ headed - we think we’d be understood anywhere - where Geordies wouldn’t’. Hence we have the comparison of Self, which is seen as the norm, with Other, which is counter-norm. Six of the eight informants who responded to the question in the negative were female. Reasons for disliking the accent tend to focus on it ‘sounding common’. Six informants were indifferent, stating that they neither liked nor disliked their accent, and three were somewhat non-committal in stating that it was ‘alright’. Variation in the evaluation of the accent across age is presented in Table 5.3 below. 203

Table 5.3: Informants’ evaluation of accent (age differences) Affirmative Non-committal Indifferent Negative Total n. % n. % n. % n. % o 8 7 87.5 0 0 1 12.5 0 0 M 8 3 37.5 0 0 1 12.5 4 50 Y 16 5 31.25 3 18.75 4 25 4 25

Seven of the eight older speakers evaluate their accent positively. The only age group with a higher proportion of its members evaluating the accent negatively is the middle group. This could be argued, again, as being indicative of age-graded behaviour, given the middle speakers’ work-oriented life stage and considering that the majority of the young speakers also claim to like their accent. However, it could also be argued to denote changing attitudes towards the accent, as none of the older informants claimed to dislike the accent, unlike four of the middle speakers and four of the young speakers.

The only group of speakers from which all members replied in the affirmative is the older males. The only group of speakers from which no members responded in the affirmative is the young adult females. Overall, more males than females and more older speakers than younger speakers claimed to like the accent.

5.1.3.2 Perceived linguistic accommodation

The second question in this section seeks to ascertain informants’ levels of awareness of linguistic accommodation and their perceived reasons for this. Overall the majority of informants claim to be aware of the fact that they adjust their speech depending on context (20 of the 32 informants). However, the females of the sample appear considerably more aware of their accommodatory behaviour, or are much more willing to admit to it at least. 13 of the 16 female speakers responded to the question in the affirmative, elaborating on their response by indicating that this depended on the perceived formality of the situation. An employment interview was repeatedly cited as an example of a formal situation in which they would shift towards a less localised accent. This 204 was particularly notable in the responses of the young adult females, all of whom respond to the question in the affirmative. Attitudes such as:

‘in interviews and stuff I try and de-pronounce my accent cos I think they’d think I’m common and not want me working for them’ emphasise the young adult female speakers’ concern with gaining employment and their desire to project what they perceive to be the ‘right’ image. In contrast, only seven of the 16 males of the sample respond to the question in the affirmative. Furthermore, of those seven some claimed to change their accent only ‘for a laugh’ or ‘for the crack of it’, whilst others claimed they would slow down their speech, but maintain, for example, that ‘I wouldn’t start pronouncing my s dead much’, or ‘go posh’. Some male speakers felt that under no circumstances would they attempt to accommodate towards a less localised accent, with attitudes such as ‘I would never dream of changing - take [pron. /tek/] me as I am’, ‘I wouldn’t dream of trying to put a posh accent on’. It seems, then, either that male speakers are less aware of accommodation towards a less localised accent in certain circumstances or, perhaps more probably, they are less willing to admit to it than are females. In view of the male responses to perceived gender differences in MbE considered earlier, it seems likely that, to these speakers, using a less localised accent is seen as a female trait and thus not one to which the males are overly keen on admitting.

Interestingly, a considerable number of informants claimed to be aware of accommodating towards a more localised accent in certain circumstances (17 of the 32 informants). Frequently this is felt to occur in anger or in argument. Amongst the middle speakers this shifting along a more-to-less localised scale is often felt to be associated with work. For example, one female claimed that her husband used a more localised accent in the workplace (steelworks) than at home, claiming that:

‘sometimes when he comes in from work he’s - he speaks more broadly and he swears more and then it takes him a while to adjust back to being I’m at home now with the family you know not at work with blokes - and he does change’. 205

In contrast, one middle aged male claimed to enjoy the shift back to a more localised accent after work (export sales), where he felt he spent most of the day ‘trying to talk in a neutral way’. He therefore claimed to ‘enjoy being with lads from the town centre and getting back into it’.

Overall, then, most speakers appear to be fully aware of moving along a more-to- less localised scale of speaking, depending on context. Female speakers are considerably more aware or more willing to admit to this shifting than males, however. The speakers who appear most cognisant of this are the young adult females, all four of whom claim that they would use both a broader accent and a less broad accent, depending on circumstance. Interestingly, the speakers who perhaps experience the most marked movement along the scale, given their place in the job market, the middle speakers, produce the greatest number of negative responses to the question. Nonetheless, the general attitudes towards MbE exhibited through the responses to this question indicate that in a wider context use of MbE is perceived to be a barrier to comprehension and to denote a low social status, and thus to be disadvantageous in the workplace. Despite this, the shift to a less localised accent is perceived to be achieved by ‘putting all the letters in’ or ‘looking towards putting the words in the right place’. The notion of changing vowel sounds, for example using a closing diphthong [ou] rather than a monophthong [o:] as a realisation of the GOAT vowel, is rejected consistently when enquired after. This point will also be considered further in section 6.1.2.

5.1.3.3 Perceived misidentification of the accent

Lying between two relatively easily identified accents of BrE, Geordie and Yorkshire, the Middlesbrough accent is not one that is readily identifiable by an outsider, as noted in section 3.1.1.4. Therefore, the third question of this section seeks to establish how informants would react to a perceived misidentification of their accent as either Geordie or Yorkshire. Informants’ responses to this question prove illuminating and their implications will be considered further in Chapter 6. 206

In all groups of speakers the most frequently given response is that a perceived misidentification as Geordie is more offensive than one as Yorkshire. Only five of the 32 informants claimed a preference for being referred to as Geordie rather than Yorkshire. Additionally, almost all informants claim to have experience of being misidentified as a Geordie, yet only one or two recall ever having been referred to as Yorkshire speakers. Responses to the question range from feelings of mild dissatisfaction with being mistakenly identified as Geordie to pronounced anger and irritation. Reasons for the dislike of the Geordie label also seem to vary across age.

Virtually all of the older speakers (seven of the eight) express a dislike of being referred to as Geordie. However, as most of the old speakers consider themselves to have Yorkshire accents, being referred to as Yorkshire is not perceived as a misidentification. Furthermore, many older informants express incomprehension at being frequently referred to as Geordie, with responses such as:

‘I’m from Yorkshire rtot Geordieland - they might as well call you a Frenchman instead of an Englishman’.

Only one of the eight older informants claimed she would prefer a Geordie identification, claiming ‘I don’t think it’s [MbE] anything like a Yorkshire accent’. The majority of older informants expressed opinions such as those revealed in the following exchange where A is an older male, E is an older female and CL is the interviewer:

A - I fell out with some Welsh people because they kept calling me Geordie CL - really? A - oh, did I? CL - and that’s [offensive] A - [it came to the stage] -well it is to me yeah cos it’s a bitter thing - I don’t mean I hate the Geordies but football and all this E - but to be called Yorkshire’s pleasing A - to be called Geordie - but Yorkshire - I never ever class myself as anything other than Yorkshire - never CL - really? E - we still think of ourselves as Yorkshire [we didn’t want to be Teesside] 207

A - [I bet if you asked 99%] if you asked 99% of Middlesbrough people what they would still like to be called and they’d tell you Yorkshire CL - really? A - yeah E - well the older ones I don’t know about the younger ones CL - I know that’s the thing there could be big age differences between them E - yeah CL - well how - you’re mostly mistaken for Geordie though aren’t you? - have you ever been mistaken for Yorkshire? - has anyone ever said you’ve got a Yorkshire accent? E - no CL - but Geordie you have? E - not so much me but [A] a few times haven’t you? A - yeah E - in Scotland he’s been mistaken they say [he’s a Geordie] A — [they call me a Geordie there] and I’ve listened to myself when I’m you know like a message say on the answering machine and there’s no way I can see we’re talking about Geordie - I mean we go away with Geordies and we’re nothing like it - but they will call you Geordie CL - do you hear yourself as Yorkshire? A - oh yeah I’d still I’m always tempted now to still put Middlesbrough, Yorkshire on my letters CL - really are you? E - well yeah I still think we should be Yorkshire A - it was the saddest day of our life when we were taken out of Yorkshire CL - really? E - well I wouldn’t say the saddest day but A - listen - well my saddest day it was - God’s country man you’ve got to get with it CL - why though? A - don’t know - proud of it CL - really? E — probably just because from being kids we were Yorkshire A — we were always Yorkshire every time everything E - I mean the thing that [A] used to say erm Yorkshire cricket it’s not the same now - but Yorkshire cricket you had to be bom in Yorkshire to play CL - yeah? E - you couldn’t play if you were born outside the borders and Middlesbrough was in Yorkshire so people from Middlesbrough could play for Yorkshire

It is clear from this exchange that an identification as Yorkshire is not, in fact, a perceived misidentification, but is how the speakers would identify themselves. 208

The middle group has the highest number of informants who claim to prefer a Geordie misidentification to a Yorkshire one (three of the eight). Positive feelings towards Yorkshire are still in evidence among some speakers from the middle group, however, with certain speakers claiming to prefer the label ‘Yorkshire’. An example opinion can be seen in:

‘if I went to Scarborough and people said you were a Geordie you’d feel that you’d want to put them right - whereas if you went to Newcastle and they called you Yorkshire you’d just let it pass probably’.

The majority of the speakers from the combined young group (13 of the 16) claim that they would object to any perceived misidentification as Geordie, with many young speakers professing a deep dislike of the Geordie accent as the reason for their offence at the term. Most young speakers recounted the experience of having been mistaken for a Geordie. One adolescent male speaker even claimed to have been mistaken for a Geordie by a group of speakers from Sunderland, which lies just 13 miles south of Newcastle and 25 miles north of Middlesbrough, perhaps indicating the growing similarity of MbE with Tyneside English. Some young informants expressed surprise at the notion of being mistaken for Yorkshire speakers, as it did not seem a realistic possibility, the Yorkshire accent being associated with both rurality and a higher social class. Responses of young speakers range from the opinion that to be mistaken for a Yorkshire speaker suggests that people ‘think you’re posher than you are because that’s a posh accent’, to the straightforward ‘I’m not a farmer’.

5.1.4 Summary of perceptions of language

The following points summarise the perceptions of and attitudes towards MbE as revealed through responses to the questions in the IdQ. • Age-correlated changes are in evidence in the defining and labelling of the accent. These changes in the labelling of the accent are in line with the changing identity of Middlesbrough in terms of local administrative boundaries. 209

• Fewer young speakers claim to be able to recognise a Middlesbrough accent than middle and old speakers. • Males are more confident of their ability to identify MbE than females. • The delimitation boundaries indicating where people are perceived to speak in a noticeably different accent are drawn closer to the south of Middlesbrough than to the north.

• Use of [?] for intervocalic /t/ and [f] and [v] for /0/ and /d/ respectively in the speech of the young as opposed to the middle and old speakers is cited as an example of perceived age-correlated variation. • The perception that adolescents speak differently from young adults is noted by young adults. • Attitudes towards female speech have changed across age - amongst old speakers, females are perceived as being less broad than males, amongst middle speakers they as perceived as being as broad as males and, amongst young speakers, females are perceived as being broader than males. • More males than females claim to like their accent. • None of the young adult females claim to like their accent. • Most speakers are aware of shifting along a more-to-less localised scale of speech, depending on context. • More females than males claim to be aware of, or are willing to admit to this accommodation. • Only five of the 32 informants state that they would not object to being referred to as Geordie. • Most informants have experienced a perceived misidentification as Geordie, whilst few had ever been referred to as Yorkshire speakers. • Many speakers profess a dislike of the Geordie accent.

The implications of many of these points will be considered further in Chapter 6. We turn now to a consideration of the informants’ perceptions of place through an analysis of their responses in the IdQ section of the interview concerned with area. 210

5.2 Place

5.2.1 Defining and delimiting place

Insight into how Middlesbrough and the area are defined and delimited is gained through responses to the following questions: • If you were watching a regional television programme, what places would you expect to hear news from? • Do you remember when the county of Teesside was formed and Middlesbrough was no longer in Yorkshire? Do you think this change made a difference? • Would you consider Teesside to be in a larger ‘north-eastern’ part of the country or a larger ‘Yorkshire’ part of the country? Why? • What do you consider the local football derby to be?

Opinions as to where regional news should reach, as elicited by the first question of this section, provide insight into the larger region of which Middlesbrough is perceived to be a part. In response to this question, apart from towns within the Tees Valley, the most frequently mentioned place is Newcastle, which is cited by 29 of the 32 informants. Interestingly, many informants claim that there is a definite Newcastle bias in the regional news, with opinions such as ‘it should be called Tyne Tyne TV mostly cos it’s very Geordie-centric’. The next most frequently mentioned place is Sunderland, which is cited by 17 informants. Carlisle and Cumbria are also cited by one or two informants. However, this is qualified by one informant with the opinion that ‘I don’t feel any affinity with Carlisle but I do with Newcastle’. As regards locations south of Middlesbrough, Whitby is mentioned, and occasional features are felt to come from York and Scarborough. One informant also mentioned Harrogate. In terms of the larger region that Middlesbrough is part of in terms of media groupings, then, the focus appears to be northwards and most particularly towards a region which includes the city of Newcastle.

The responses to the second question concerning the formation of Teesside and the removal of Middlesbrough from North Yorkshire are illuminating and are 211

connected to the question considered in section 5.1.3.3. By far the majority of the old speakers express regret at no longer being part of the Ridings of Yorkshire, with opinions such as those in the exchange given in section 5.1.3.3 in evidence. In aligning themselves with Yorkshire, some of the older speakers see themselves in direct opposition to locations further north, for example:

‘what I always say — we are the Yorkshiremen trying to keep the Geordies out of Yorkshire - cos we’re on the borderlines you see - and all these people, Guisborough and all that -we have to defend the front line you see’.

Not all informants held the same opinion, however. One older female claimed: ‘I never felt we were in Yorkshire because Stockton was in Durham then wasn’t it? in those days - so it seemed silly Middlesbrough in Yorkshire - it was just the river wasn’t it? that was the boundary - so I never felt we were in Yorkshire even though we had to put it.’

In the middle group, the majority of speakers express the opinion that the conurbation both north and south of the River Tees should be brought together and not separated by the river. Three of the speakers claim that Middlesbrough should go back to Yorkshire, however. Amongst the middle group there is considerable expression of a lack of identity in Middlesbrough, with opinions such as: ‘we’re not Geordie - we’re not Yorkshire - we’re nothing really’

‘we’re no-man’s land, aren’t we? - we don’t know what we are’

‘I remember people saying things like Geordies won’t have you and Yorkshire won’t have you and all that - as if we were almost sort of nothing really’

‘I don’t feel we’ve got an identity I don’t know why’

‘we haven’t got a status at all - they don’t know what we are’.

In the combined young group a much greater majority of speakers express the opinion that the conurbation around the Tees should be grouped together (13 out of 16). Two young speakers claimed that Middlesbrough should be a city by itself and only one expressed a desire to be in Yorkshire. A few of the young 212

speakers were unaware of the association with Yorkshire and so appeared puzzled by the question. Responses, then, correspond exactly to the fact that the speakers from the old group have spent the majority of their lives in Middlesbrough, Yorkshire; the middle group’s lifetime has seen constant changes of identity; and the young group have no memory, or in some cases no knowledge, of the Yorkshire identity.

In terms of local allegiances, the importance of popular culture cannot be overstated, and the significance of sport in popular culture appears central to many informants’ sense of identity. Rivalry, particularly football rivalry, is often cited as a reason for the hostility towards Geordies amongst the young speakers, with no sense of rivalry felt towards football teams from Yorkshire. Although the rivalry is expressed as open hostility, rivalry also suggests a relationship. Where no rivalry is felt, no relationship can be seen to be in existence. Speakers from the old and middle groups who have strong feelings for Yorkshire often express their allegiance in terms of cricket, and the former eligibility to play for Yorkshire Cricket Club is seen as a valid and emotive reason for wanting to be connected to Yorkshire, as noted in the exchange produced in section 5.1.3.3. The declining popularity of cricket and the increasing popularity of football in recent years, may also have had a part to play in the shifting allegiance and sense of identity. Additionally, the view that an urban centre becomes prominent nationally when it has a football team in the Premier League in England is expressed repeatedly by speakers. The fact that Middlesbrough Football Club has had a particularly high profile in the media since 1994, with the arrival of internationally renowned personalities at the club, is commented on frequently by informants and may also have contributed to the growing confidence of the young people in the status o f Middlesbrough as a place in its own right on a national scale.

Interestingly, in light of the responses to the second question of this section, in response to the third question, that of whether informants would consider Middlesbrough to be in a larger ‘North Eastern' part of the country or a larger ‘Yorkshire’ part of the country, all of the informants of the sample answered ‘North Eastern’. Some of the old and the middle speakers considered the 213

‘Yorkshire’ option, but all decided on ‘North East’. Even the speakers who considered themselves to be from Yorkshire gave responses such as ‘you’re right at the top of Yorkshire so to get it across you’d have to say North East’. Some informants had never really considered the proximity of Yorkshire before. The idea of classing Middlesbrough with Yorkshire in a geographical sense appeared strange, with responses such as:

‘it’s weird, even though you’re the same distance, how much you don’t class yourself with them’

‘it’s weird when you only go two minutes down the road and you’re in like North Yorkshire - no I don’t consider it at all - no I would definitely not say it [Yorkshire]’.

In view of the unanimous answer of ‘North East’, then, it seems either that the geographical area of the locality and the actual identity of the region are not one and the same, or that speakers, even those who still consider themselves to be Yorkshire people with Yorkshire accents, now accept the label of North East as a response to the question ‘where is Middlesbrough?’.

The definition of a local derby is ‘a football match between two teams from the same area’ (Collins English Dictionary). The informants’ opinions of what constitutes a derby match with Middlesbrough Football Club also provide insight into which area Middlesbrough is thought to be located. In response to this question, the vast majority of informants specified a match between Newcastle United Football Club or Sunderland Football Club (some informants were unsure as to what a derby match was and so did not provide an answer). An equal number of informants mentioned Newcastle and Sunderland (26). However, 13 informants stress that Newcastle United would be the more important opponent in terms of local rivalry. Six informants mention Leeds United also. However, Leeds is always mentioned after Newcastle and Sunderland and in terms such as ‘it used to be Leeds’, ‘Leeds to a lesser extent’. A derby match with Leeds United is also thought by one informant to be ‘something which has been built up through all the fighting and aggression - not really a local derby - more of a hooligan aspect’. 214

Overall, then, the region which Middlesbrough is perceived to be part of is the North East rather than Yorkshire. Although strong feelings for Yorkshire and regret at no longer being considered part of the North Riding of Yorkshire are expressed by the majority of the older speakers, Middlesbrough, as a place, can be seen to be defined as North Eastern. Furthermore, the region of the North East can be seen as being delimited by the urban centre of Middlesbrough at its southernmost tip.

5.2.2 Attitude towards place

An idea of general attitudes towards Middlesbrough is provided by the responses to the following questions: • If you could, would you change where you come from? • What do you consider the best and worst things are about growing up and living in Middlesbrough? • If an outsider was complaining about Middlesbrough, would you defend it even if you agreed with what s/he was saying? Why/why not?

The responses to the first question of this section suggest that, in general, positive attitudes towards Middlesbrough are felt by the informants, as 28 of the 32 replied that they had no desire to change their origins. Three responded in the affirmative, however. All three were young speakers. Furthermore, two of the three were young adult females. Although the other young adult females responded affirmatively, the responses were not wholly positive, for example:

‘this is where I’m from - it’s where your family are and where you grew up - but I don’t want to live here the rest of my life - as soon as I can I’m out of this place’.

The responses to the second question in this section give an impression of what particular problems the urban centre of Middlesbrough is perceived to have by its inhabitants as well as any benefits of it as a place to live. Table 5.4 shows the features mentioned as the most beneficial aspects of Middlesbrough in order of frequency of citation. 215

Table 5.4: Perceived advantages of life in Middlesbrough

Feature Number of times mentioned Location 15 People 11 Nightlife 5 Shopping 3 Football team 1 Weather 1 Steel (employment) 1

In terms of the most beneficial aspect of Middlesbrough, somewhat telling is the fact that the most frequently given response is not a feature of Middlesbrough itself. Rather it concerns the geographical location of Middlesbrough and more particularly locations which can be easily reached from Middlesbrough. 15 informants’ initial response mentioned location as the most advantageous feature, with access to the coast, the moors, and the hills discussed. Table 5.5 below presents features perceived to be the least beneficial aspect of life in the town.

Table 5.5: Perceived drawbacks of life in Middlesbrough

Feature Number of times mentioned Pollution 11 Crime 10 Unemployment/lack of prospects 5 Nothing to do/lack of facilities 4 Drugs 2 Reputation/image 2 People 2 Estates 1 Weather 1 Narrow-mindedness 1 Public transport 1 216

The most frequently mentioned feature perceived to be disadvantageous is the level of pollution from industry. This is not altogether surprising given the frequency of the mention of the surrounding countryside as the most auspicious feature of the urban centre, as noted in Table 5.4.

The final question of this section concerns the level of instinctive community allegiance. Overall, the majority of informants claimed that they would speak in defence of the town even if they were in agreement with the complainer (17 of the 32). Nine informants maintained that they would not defend the town and six stated that it would depend on whether they agreed with the complaint or not. The males of the sample were found to be more loyal to their town in terms of the question posed than the females (11 of the 16 males answered in the affirmative, and only three responded in the negative compared with six of the 16 females responding positively and six negatively). Additionally, the older speakers were revealed to be more willing than the middle or young speakers to defend the town against criticism, whether warranted or not. Six of the eight older speakers responded in the affirmative compared with four of the eight middle speakers and seven of the 16 young speakers.

Six of the informants likened the defence of place, speaking in favour of Middlesbrough if under criticism, to how they would feel if their family were being criticised, with opinions such as:

‘yes — no question - because it’s our birthplace and we’re quite proud of it - it’s like if someone was saying something about my sons’.

The similarity also lies in the perception that, as in a family, only an in-group member has the ‘right’ to criticise another: for example, ‘the only person who has a right to slag off Middlesbrough is someone who lives there’. The extent to which place forms part of the individual’s identity is also revealed in responses such as, ‘if you slag Middlesbrough off you’re slagging me off cos 1 live there’.

Overall, then, the attitude towards Middlesbrough as revealed in the responses to the questions of this section is positive. The vast majority of informants would 217

not change where they come from. That is, they are not dissatisfied with Middlesbrough as the place of origin component of their identity. Also the majority of informants feel a degree of in-group loyalty which manifests itself in a desire to counter criticism of the town when this originates from an outsider.

5.2.3 Perceptions of image

The questions which seek to obtain insight into the perceived image that Middlesbrough has, both in the minds of inhabitants and in the media, and therefore in the minds of outsiders, are: • What image or description of Middlesbrough would you give to someone who didn’t know it? • Have you ever seen Middlesbrough on a national T.V. programme (e.g. a documentary)? If so, how was it portrayed? • Do you think Middlesbrough is a fashionable place to be?

By far the majority of informants claimed that an industrial image would best represent Middlesbrough to an outsider. 20 of the 32 informants of the sample claimed they would describe Middlesbrough in terms of its industry in order to project a representative image of the place to an outsider. Some informants, particularly older ones, stressed that the heavy industry is no longer a prominent feature of the town, for example:

‘what is there here? - there’s nothing here - the image that people had of Middlesbrough years ago was the steel - cos that’s what it was - it had a name - ‘Steelborough’.

The image of heavy industry persists, however, with some younger informants mentioning that the image of industry was one taught to them at their educational establishments.

The town’s geographical location, as considered in the previous section, is also evoked as an image of Middlesbrough in the minds of the informants. Ten informants mentioned the close proximity of country scenery. The nightlife, also, was mentioned by informants across age as a feature which represents the town. 218

Overall, 16 of the 32 informants claimed they would give a positive image of the town to an outsider. Eight said they would give a neutral image. Eight declared that they would give a negative image, however, with opinions such as, ‘it’s a miserable grey town full of miserable grey people’.

Responses to the second question provide insight into how informants perceive Middlesbrough to be represented through the media. Any representation of the town through the media would contribute to the image held by outsiders. The vast majority of informants, 27 of the 32, had the impression that Middlesbrough is represented negatively in the media. This is seen to be both through television and through national newspapers. Economically, this is felt to have an adverse effect, as it is perceived to be off-putting to people wishing to come to work in the area. Middlesbrough is felt to get the ‘outside toilet, flat cap treatment’, or be given an ‘Andy Capp image’, or be associated with ‘working men’s clubs and whippets’. Interestingly, given the responses to the previous question, a major criticism is that industry is always shown to represent the locality, regardless of the topic of the feature in question. A tension is revealed, then, between the evocation of an industrial image by the informants and its effects when used by the media to represent the town. In the same way that, to some informants, it is felt that an insider can criticise Middlesbrough but an outsider has not earned the right, it seems that an insider can use an industrial image to represent the town but an outsider similarly does not have the right. The image of industry which surrounds Middlesbrough, then, appears to be simultaneously accepted and resented.

In response to the third question of this section concerning whether or not Middlesbrough is a fashionable place to be, the majority of informants appeared unsure. Of the young speakers, only young adult males responded without hesitation in the affirmative (three of the four). Clothes shops and nightlife were discussed with reference to the question. All other young speakers either responded negatively or were uncertain. This finding appears less surprising when considering that nine of the 16 young speakers claimed that the retail outlets of the town, particularly those selling designer clothing, cater more for male consumers than female. Furthermore, males are considered to be more 219 willing to spend money on clothes and are thought to be more interested in image than females, which was an opinion expressed by six of the eight young female informants. Of the adolescent males, interest was also revealed in designer clothing. However, the type of attire under consideration, that from established designer houses, was considered ‘too expensive for a 17 year old’. Three of the middle speakers and two of the older females also claimed that Middlesbrough was a fashionable place. Interestingly, most of those who responded in the affirmative stressed that the town had become more fashionable in recent years, with most specifying the last five years as a period during which the status of Middlesbrough had witnessed a change. These findings will be explored further in the following chapter.

5.2.4 Summary of perceptions of place

• In terms of popular culture, Middlesbrough is perceived to be in a region which includes urban centres to the north, specifically Newcastle. This is manifested through regional television groupings and football derby matches. • In a geographical sense, Middlesbrough is unanimously felt to be part of the North East. • In terms of a local government regional identity, the older speakers regret no longer being a part of Yorkshire, whilst the middle and younger speakers claim that the conurbation along both sides of the River Tees should be together and should have its own identity. • The middle group express a considerable sense of a lack of identity. • A lack of relevance of the label ‘Yorkshire’ is found amongst the young speakers. • Sport appears central to local allegiance for many informants. Sport, particularly football, also appears significant in the growing confidence amongst young speakers in Middlesbrough as a place in its own right. • Overall, positive opinions are expressed by informants towards place. • The most beneficial aspect of Middlesbrough is felt to be its geographical location. • The least beneficial aspect is the pollution from industry. 220

• A tension is revealed between an insider’s use of an industrial image to represent Middlesbrough and that of an outsider. When used by an outsider, specifically through the media, an industrial image is felt to be unwarranted, and harmful to the prospects of the community. • The majority of informants claim that Middlesbrough is represented negatively through the media. • Young adult males perceive the town to be a fashionable place in terms of clothing and nightlife. • Young females perceive young males in Middlesbrough to be more concerned with image than are females. • The status of the town on a national scale is perceived to have changed in recent years.

5.3 Strength of local allegiance

In order to gain some awareness of the level of feeling of local allegiance amongst the informants and to reveal whether differing levels correlate with the age and gender of the speaker, two sources of information are utilised. First of all, a general impression of how the informants responded to the questions posed in the IdQ part of the interview will be recorded. Secondly, results from the Affiliation Score Index will be presented.

Although the responses to all questions posed in the IdQ are borne in mind, particular attention is paid to those concerning the question of whether informants like their accents and of whether they would give a positive, negative or neutral image of Middlesbrough to an outsider. Reactions to the questions in the IdQ are categorised as points along a cline from negative to positive, with five points identified: negative, between negative and neutral, neutral, between neutral and positive, and positive. The results for the four individual speakers of each speaker group are presented in Table 5.6 below. 221

Table 5.6: Levels of local allegiance for individual speakers as revealed through responses to IdQ

Negative Neutral Positive

Adm VV vv

YAm vvvv

Mm V V vv

Om V vvv

A df vv V V

YAf V vv

M f vv V V

O f V vvv

Levels of local allegiance are revealed to be correlated with age and gender variables as presented in Table 5.6. Males demonstrate considerably higher levels of local allegiance than females of the sample. Of the 16 male speakers 11 are located on the most positive point of the cline. Four of the remaining five appear in the neutral and neutral to positive points, with only one male responding negatively in the interview to language and place. The female speakers are much more evenly distributed across the cline. Two females reacted neutrally to the questions on the IdQ and, of the remaining 14, seven are located in the left hand side, neutral to negative, whilst seven appear in the right hand side, neutral to positive. Across age, the older speakers appear more positive than the other age groups, six of the eight older speakers being placed in the positive point. The middle and younger speakers are fairly equally placed, half of both the middle speakers and the younger speakers being placed in the neutral-to-positive section. Of the individual speaker groups, the most positive is the young adult male group, in which all four speakers responded positively to their language and town community. The least positive responses appear to be in the young adult female and the middle female groups, where negative responses were revealed amongst five of these eight speakers. 222

The assessment of the strength of local affiliation based on the responses to the IdQ is necessarily a subjective judgement. In order to counteract this subjectivity and provide a more objective element to the rating of levels of local allegiance, the results from the Affiliation Score Index are also presented. The Affiliation Score Index (presented as Appendix 6) was administered after the interview and is designed to test in-group preference, as outlined in section 3.2.3.3. The Affiliation Score Index consists of seven multiple-choice questions which test the strength of the informant’s local affiliation. The responses to the questions carry a score of 1, 2 or 3, where 3 indicates the highest level of local affiliation. Therefore, a minimum score of 7 and a maximum score of 21 can be reached. The mean scores of the eight speaker groups are presented below. As in the reaction to the IdQ, scores in the Affiliation Score Index can be seen to represent points on a cline, where the lower the score the more negative the response, and the less strong the informant’s feeling of local affiliation and the higher the score the more positive the response and the stronger the feeling of local affiliation. Broadly, a score from 7-14 indicates a negative to neutral response, and a score from 15-21 indicates a neutral to positive response. Whether the patterns found in the responses to the IdQ are duplicated in the more objective Affiliation Score Index is considered below, as Table 5.7 presents the mean group scores for all eight speaker groups.

Table 5.7: Mean group scores for Affiliation Score Index

Speaker group Mean score Adm 14.75 YAm 18.25 Mm 16.75 Om 15.00

Adf 15.75 YAf 12.25 M f 16.00 O f 17.00 223

As can be seen, the results from the Affiliation Score Index correspond to the reactions to the IdQ in that the speaker group with the highest score, and therefore the strongest feeling of local affiliation, is the young adult male group (the group from which all members demonstrated a positive reaction to the questions posed in the IdQ). The group with the lowest score and therefore the least strong feeling of local affiliation is the young adult female group (the group from which no members responded positively to the questions in the IdQ). The young adult females are the only speakers of the sample who responded negatively, in that they are the only group with a mean score of below 14. The majority of informants achieved a positive rating on the Affiliation Score Index, and therefore demonstrated a high sense of local allegiance in terms of in-group preference and a feeling• of being emotionally tied to the area. 2

5.3.1 Summary of strength of local affiliation

• Overall, male speakers demonstrate a higher level of local affiliation than female speakers. • Overall, older speakers demonstrate a higher level of local affiliation than middle or younger speakers. • All young adult males respond positively to their language and community, and the highest group mean score in the Affiliation Score Index is recorded from this group. • The lowest group mean score is found in the young adult female group, and none of the four members of this group respond positively to their language and community.

The data presented in this chapter have revealed the multi-layered construction of identity. Age- and gender-correlated variation is in evidence in the responses. From the attitudinal data, we see both what larger community the informants believe theirs to be a part of and what smaller cohort identities are perceived to

2An interesting finding from the Affiliation Score Index was that in response to question 4 which questions the preferred accent of the informant’s child’s school teacher, of the 32 informants only one responded that he would prefer the teacher to be a local person with a local accent. 224 exist within the town community. We now turn, in Chapter 6, to consider how the attitudinal and perceptual findings can inform the linguistic trends which were presented in Chapter 4. Chapter Six Synthesis and Discussion

6.0 Introduction

In this chapter the linguistic trends observed in the data and presented in Chapter 4 will be interpreted in light of the attitudinal information procured through the method of data elicitation used in the study and described in Chapter 5. The motivation for the linguistic variation, and the change in progress in MbE it suggests, can be explored in more depth by attempting to access the local knowledge that speakers operate with when constructing their collective and cohort identities, and by assessing how forms are evaluated and what symbolic function they perform. Thus the attitudinal information obtained can be correlated with and can inform the linguistic evidence uncovered.

The linguistic trends observed in the data include the increased use of localised features, suggesting the convergence of MbE with varieties further north. Also the adoption of innovatory forms was revealed, with forms which are believed to be spreading nationally from a South Eastern epicentre present in the data along with innovatory forms which are reported in Newcastle English. Certain of the innovatory forms were found to have a restricted use in the speech of sub-sets of the young of the sample, whilst others have been adopted widely by the young speakers. Speaker-based motivational factors which may elucidate these trends will now be considered.

6.1 Ideology

By utilising an ethnographic approach to the sociolinguistic enquiry into MbE and by gaining access to speakers’ opinions and overt discussion and awareness of what their language and area mean to them, we gain insight into the symbolic function of the dialect. A language ideology framework allows us a deeper understanding of the overt and covert social meaning attached to the linguistic 226 forms under consideration, as outlined in Chapter 2. Such a language ideology framework, based on linguistic anthropological theories (see further Silverstein 1992, 1995, Woolard 1992) has recently been applied to sociolinguistic data (see further Anderson and Milroy 1999, Dyer 2000, Milroy 2000, 2001). If we take a model of language ideology as a neutral semiotic process (Silverstein 1992, 1995), as opposed to a strategy for maintaining power in an institutional sense (Lippi-Green 1997), we can consider how language forms index the social identity of the speaker. Silverstein (1995) identifies two orders of indexicality, as noted in section 2.3.1: first order indexicality, the association of a linguistic form with a socially meaningful category; and second order indexicality, the overt or covert noticing, discussion and rationalisation of first order indexicality. The opinions expressed by the informants of the study, as detailed in Chapter 5, are thus overt instantiations of second order indexicality. Observations about language, beliefs about community, and evidence of style shifting can all reveal ideologies underpinning the informants’ choices of phonological variants. Examination of these instantiations of second order indexicality can thus offer a more detailed explication of linguistic variation than a simple correlation analysis. We begin with an examination of the motivation for the apparent convergence of MbE with varieties further north in section 6.1.1. We then consider how variants are evaluated in the community, including the informants’ awareness of linguistic change in progress and the extent of style shifting in the data, in section 6.1.2.

6.1.1 Local orientation and linguistic variation

A major focus of the study is the identity of the locality under investigation and, more particularly, the fluid nature of regional identity construction. The problematising of issues of identity has been a feature of recent sociological work on the English town of Berwick-upon-Tweed, which lies very close to the Scottish border (Kiely, McCrone, Bechhofer and Stewart (2000)). The specific identity construction of localism was found to be employed by the ‘Berwickers’ which prioritises the ‘Berwicker’ identity over the national and regional identities available. This strategy was more favourably viewed in some contexts than others, however, with inhabitants of nearby towns north and south of the border 227

(respectively Eyemouth and Alnwick) rejecting the strategy of localism as used by the ‘Berwickers’. The interest of the present study of MbE is not in national identity, but rather in regional identity and the strategies used by informants to define and delimit their identity in terms of the region, rather than the town, in which they live. The relativity of the concept of community and the importance of larger boundary relationships through which identity is constructed differentially are therefore of interest. Linguistic trends observable in the data are thus seen in relation to the identity strategies utilised by informants and their sense of the identity of Middlesbrough.

The concept of place in terms of the identification of speech communities as well as the conceptualisation of space and the delineation of region have been considered in section 2.1.1. In the past, the construction of place has largely been unexamined and treated as a given in variationist studies. However, the importance of the psychological reality of where speakers believe themselves to be from appears central to the identification of such concepts as ‘speech community’. Although speakers can be seen as being products of geographical place, it is the speakers who construct place. Johnstone (1999: 8) argues ‘[rjegions have come to be seen as meaningful places which individuals construct as well as select as reference points. Identification with a region is identification with one kind of ‘imagined community”. Importantly, the psychological reality of place can shift, and the construction of community identity can be in a state of flux.

Combined with having something of a ‘border town’ status, which can entail a problematic construction of identity as noted in Berwick-upon-Tweed, the complexity of ascertaining the identity of Middlesbrough is further compounded by the fact that it has witnessed a recent shift in identity, as outlined in section 3.1.1.4. The pulling of Middlesbrough out of Yorkshire and into the North East correlates neatly with the convergent linguistic trend suggested by the higher level of use of glottalised stops, most particularly [?p]. Evidence suggests that the incidence of glottalisation of word-medial (p t k) is markedly higher in Tyneside English than in MbE. Therefore, in increasing use of word-medial 228 intervocalic glottalised stops the young speakers of the sample are bringing MbE closer to Tyneside English. In this light, the motivation for the sudden increase in the use of a localised North Eastern feature may seem straightforward. We could hypothesise that the shifting identity of Middlesbrough has resulted in inhabitants now identifying with varieties of English found further north, most particularly with the variety of the dominant urban centre of the North East, Newcastle. Consequently, they are increasing their use of a North Eastern feature. However, compelling as this deduction may seem, this can only be conjecture, and the motivation for the apparent convergence must be addressed in more detail by assessing speakers’ local orientations.

In terms of physical local orientation, there may be evidence for increased short­ term contact between people from Middlesbrough and people from Tyneside. Improved roads and public transport systems have made travel between the urban centres relatively easy (by road, almost the same time is needed to travel by public transport from Middlesbrough to Newcastle (38 miles) as from Middlesbrough to Hartlepool (an urban centre 15 miles away on the north bank of the mouth of the Tees)). Indeed, ten of the 16 young speakers in the sample claimed that Newcastle would be their usual destination for a day trip, as compared with four of the eight speakers from the middle group and one of the eight speakers from the old group1.

This suggested increase in face-to-face contact with speakers in Tyneside is combined with the potential influence exerted through increased exposure to and awareness of accents from further north as presented through the media. Since January 1959 Middlesbrough has been included in the independent regional television grouping of Tyne Tees, as noted in section 3.1. As a large part of regional television broadcasts concerns local news or features, this involves the interviewing of ‘local’ people with ‘local’ accents. Although presenters and newscasters do not generally speak with marked regional accents, the appearance

1 The question ‘If you wanted a day out shopping, where would you go?’ is included in the interview, as noted in section 3.2.2.6, to ascertain in which nearby urban centre speakers from Middlesbrough would be most likely to experience short-term face-to-face contact with speakers o f other varieties o f BrE. 229 of ‘local’ people on regional television programmes has had the effect of exposing the inhabitants of Middlesbrough for the past 40 years to varieties of English which are spoken in communities further north. Indeed, certain speakers expressed the view that the inclusion of Middlesbrough in the regional television grouping of Tyne Tees has led to the accent becoming closer to Geordie. Added to this is the recent popularity, on a nation-wide scale, of a number of television presenters and celebrities with North Eastern accents who appear on national television programmes aimed largely at a youth market, such as Anthony McPartlin and Declan Donnelly, Jayne Middlemiss, Robson Green and Donna Air. Although no empirical evidence exists to determine the effect of the media on linguistic behaviour, it has recently been suggested that:

speakers of varieties which are linguistically close to socially influential models would be more likely to adopt patterns from those models than speakers whose native forms are very different. This ought to hold true for media-influenced change just as it does in the case of face-to-face accommodation (Foulkes & Docherty in press).

Following this argument, if North Eastern accents are becoming more socially influential (particularly in the youth market) as would be suggested by their increased use in the media, speakers from Teesside, being linguistically close to the model, would be those most inclined to adopt patterns from the model.

Although contact with and exposure to accents from further north may have increased, contact alone does not explain the motivation for the linguistic changes. We cannot simply assume that speakers from Middlesbrough identify positively with varieties of English found further north, in particular the accent of the major urban centre of the North East, Newcastle, and that this positive identification with Newcastle is the motivation for the increase in glottalisation. We must question how the speakers perceive themselves and with which groups they identify positively.

The attitudinal information presented in Chapter 5 reveals clear generational differences in linguistic orientation in Middlesbrough, as speakers’ self-images, in terms of what accent they perceive themselves to have, differ across age. The 230 way speakers overtly define and delimit their accents is central to the construction of community identity. Just as ‘popular labels for places often reflect the ways in which places are constituted through shared experiences and shared orientations’ Johnstone (1999: 7), so the labels people choose to apply to themselves can reflect shared orientations. In claiming to have a Yorkshire accent, speakers from the old group in the sample are demonstrating a shared orientation towards Yorkshire. Similarly, speakers from the middle group demonstrate a shared orientation towards Teesside, whilst young speakers display a shared orientation towards Middlesbrough.

As well as the definitions of Self that the speakers operate with, however, the labels which are applied to the speakers by outsiders may be instrumental in the construction of these shared orientations and identities. The extent of the use of the label ‘Geordie’ as a perceived misidentification of the informants was noted in section 5.1.3. The influential role of labelling has been investigated widely in sociological analyses of deviance (see further Gove 1980). Part of the labelling theory may be applied to the picture of language variation under consideration. One of the most influential statements on deviance by Becker (1963: 9) contains the following tenet: ‘[t]he deviant is one to whom the label has been successfully applied; deviant behavior is behavior that people so label’. In other words, an act only becomes deviant when others perceive and define it as such. From this perspective, we can argue that an accent will become North Eastern when others perceive and define it as such. Becker goes on to argue that, as an individual’s self-concept is largely derived from the responses of others, s/he tends to see her/himself in terms of the label. This may produce a self-fulfilling prophecy whereby the deviant identification becomes the controlling one. Given the extent of the application of the label ‘Geordie’ to the informants in the Middlesbrough sample, as noted in section 5.1.3, we could suggest that inhabitants of Middlesbrough are influenced subconsciously by the label ‘Geordie’ and not by the label ‘Yorkshire’ in the construction of their identities. With the redrawing of administrative boundaries and the subsequent shifting identity of Middlesbrough, the label ‘Yorkshire’ has lost all relevance for the young speakers in the town, for whom no connection is seen to exist between Middlesbrough and Yorkshire (recall that older speakers consider themselves to be Yorkshire people with 231

Yorkshire accents). Given this dissociation with Yorkshire, perhaps the ascribed status has become the achieved status, and the deviant identification of Geordie has become the controlling one. Thus, a self-fulfilling prophecy may have been produced whereby MbE has become more North Eastern, demonstrated in the fact that young speakers are increasing their use of a localised North Eastern feature. Nonetheless, this still does not disclose whether or not a positive identification with the Geordie accent and the dominant urban centre, Newcastle, is felt by the speakers who are instrumental in effecting the convergent linguistic trend.

The instantiations of second order indexicality presented in Chapter 5 reveal the realignment of orientations across age groups and denote the ideologies underpinning the speakers’ choices of sociolinguistic variants. Middlesbrough, clearly, is a different place for the different generations in the sample. The increased use of glottalisation can thus be seen as being ideologically motivated, as it appears to be contingent on local images of salient social categories that have shifted over time. Furthermore, the data presented in Chapter 5 reveal how perceived space is crucial in identifying these salient local groups. The significance of the North East, and the irrelevance of Yorkshire to the younger speakers, despite its Euclidean distance from Middlesbrough, is clear in the data. (Recall the observations ‘it’s weird, even though you’re the same distance, how much you don’t class yourself with them [Yorkshire]’; ‘it’s weird when you only go two minutes down the road and you’re in like North Yorkshire - no I don’t consider it at all’.) Yorkshire, clearly, is largely irrelevant to the construction of identity of the younger speakers of the sample.

Despite this, the instantiations of second order indexicality appear to refute the hypothesis that the increased use of glottalisation demonstrates a conscious identification or an ‘act of identity’ (Le Page & Tabouret-Keller 1985) with Newcastle and Tyneside English. An ardent sense of rivalry and even hostility towards the Geordie accent and what it is perceived to stand for is demonstrated in the responses. This hostility is expressed by all four adolescent females of the sample (the adolescent females being the speakers with the most marked increase in the use of glottalised stops, and therefore those at the vanguard of the 232

convergent trend). The rivalry and hostility are expressed largely as a dislike of the accent and a resentment towards the perceived dominance of Newcastle in the North East. This open hostility would suggest that, on a conscious level, the young speakers from Middlesbrough do not identify openly and positively with Tyneside or the Geordie accent.

Instead, the higher level of use of globalised stops is concurrent with an increased confidence expressed by young speakers in the status of Middlesbrough both in terms of its accent and as a ‘place’ in its own right. The hostility towards Newcastle and the Geordie accent, and the refusal of young speakers to see Middlesbrough as a satellite of the dominant Tyneside conurbation, suggests they see themselves as ‘North Eastern’, but resolutely and specifically as from Middlesbrough. The most plausible interpretation of the increased use of globalised stops seems to be that in increasing their use of a localised feature, young speakers from Middlesbrough are not identifying with Newcastle, but are indexing their North Eastern, specifically Middlesbrough, identity. Although the question of the identity of Middlesbrough is still a matter for debate, for the young speakers it is not part of a larger Yorkshire region, as it is for the old speakers, nor is it a place without an identity, as it is for the middle speakers. Rather, for the young speakers of the sample the identity of Middlesbrough is simply Middlesbrough. As in Berwick-upon-Tweed, the young speakers of the present study are using the strategy of localism to construct their community identity, and one way of indexing this identity linguistically may be by demonstrating a higher level of use of intervocalic glottalisation. Hence, although linguistic convergence appears superficially to indicate speakers’ high regard for and identification with the possible donor variety, speakers are able to reallocate forms to index locally relevant identities. Moreover, this may occur when the groups are in tension, as Woolard and Schieffelin argue:

[cjommunities not only evaluate but may appropriate some part of the linguistic resources of groups with whom they are in contact and in tension, refiguring and incorporating linguistic structures in ways that reveal linguistic and social ideologies (Woolard & Schieffelin 1994: 62). 233

As in Dyer’s study of Corby in which historically Scottish features were argued to have been reallocated to function as indicators of local rather than Scottish identity (2000: 164), we find in MbE the recycling of a traditional feature which was revealed to be recessive in Newcastle (Docherty et al. 1997). The feature, glottalisation of voiceless stops, most particularly use of [?p], clearly has strong symbolic value, hence its increased use. However, in contrast with what may be assumed without recourse to the local knowledge and orientation that the speakers concerned operate with, it appears not to symbolise contemporary Newcastle to the speakers of Middlesbrough, but rather to symbolise something which is more locally relevant.

The role of local orientation and identification with geographical place can likewise be considered in terms of the adoption of innovatory forms in the data. In Chapter 4 we observed the very restricted use of the fronted variants of (th) and (dh) in the sample, [f] and [v] realisations of (th) and (dh) respectively were found to be overwhelmingly associated with the speech of the young adult males. The fronted variants were avoided virtually categorically by the young females of the sample. The young adult males’ use of [f] and [v] is combined with the fact that they have experienced a considerable amount of recent geographical mobility. Thus, without recourse to local knowledge, we could hypothesise that the young adult males of the sample are less locally oriented and demonstrate lower levels of local allegiance than other speakers. Rather, they are outward- looking and open to the adoption of features associated with other geographical localities. Young females, on the other hand, with their considerable increase in the use of glottalised (p) and their adoption of pre-aspirated pre-pausal and turn- final (t), both of which are features associated with Tyneside English, may be the speakers of the sample who demonstrate the highest levels of local allegiance and orientation towards the North East.

This is not the case, however. Paradoxically, the young adult males, who demonstrate the highest level of usage of assumed ‘Southern’ forms and who have experienced the highest level of recent geographical mobility, exhibit the highest level of local allegiance of all speakers of the sample. All four of the 234

young adult males responded positively to the IdQ questions in the interview. Moreover, the highest mean group score in the Affiliation Score Index, a score of 18.25, indicating the highest level of local affiliation, was recorded from the young adult male group. A similar finding was also uncovered in the recent study of Corby, where young males were found to ‘have more of a sense of local belonging than young women’ (Dyer 2000: 160). In the present study, the young adult females were found to have the lowest level of local allegiance and local orientation in the sample. All young adult females responded negatively or neutrally to the questions of the IdQ. Furthermore, the lowest score of the sample, a score of 12.25, was recorded from this group in the Affiliation Score Index. The adoption of innovatory forms appears to have little to do with positive identification with a geographical locality with which the features are argued to be associated. The use of features which are thought to be associated with the South of England clearly does not compromise the speaker’s self-identity as Northern or North Eastern, as the speakers with the strongest feelings of local allegiance and orientation are precisely those who make use of the forms in question. Likewise, increased use of localised variants does not equate with a heightened sense of local allegiance and an increased local orientation, as speakers who show lower levels of local allegiance are those who exhibit a more marked increase in the level of use of supra-local forms.

Increased geographical mobility may indeed lead to a break-up of WC territorially-based close-knit networks. However, short-term mobility, in this case at least, does not lead to a weakening of local orientation and allegiance. On the contrary, it may lead to a heightened sense of identity and identification with the home locale. This may lead to a subsequent increase in or exaggeration of localised forms (recall that the young adult males demonstrated the highest levels of use of word-medial glottalised (p) and (k) in the whole sample). It seems that there is not a linear relationship between identification with a perceived donor variety and increased use of a form associated with that donor variety. What the innovatory forms symbolise to the speakers who adopt them appears not to be related to positive identification with a perceived donor variety in a geographical sense. Rather, speakers are able to adopt innovatory forms which symbolise something of value to them without compromising their local orientation. 235

6.1.2 Evaluation of variants and linguistic variation

How forms are evaluated by speakers is problematic to ascertain, but it is clearly crucial to an attempt to understand the motivation for the adoption or increased use of certain forms. The findings detailed in Chapter 4 demonstrate that the variants of the study which are increasing or innovatory have differing distributions in the sample. Certain forms are used widely by the young speakers, whereas a more restricted use is revealed for other variants. Insight into why this should be can be gained through observation of which forms are noticed in the speech community and how they are evaluated, combined with investigation of the extent of the use of the innovatory forms in the more formal word list style of speech.

A major factor in the differing distributions of the innovatory and increasing forms of (r), (th) (dh), and (t); [u], [f], [v] and [?], [ht] respectively, could lie in the degree of salience and overt stigmatisation that the variants carry (cf. Trudgill 1986: 11). To some extent, this is indicated by the degree to which the features are noticed in the speech community. The phonemic merger involved in (th)/(dh) fronting is crucial to its salience and acceptability. By contrast, the phonetic differences involved in the use of [u] for (r) and [ht] for pre-pausal (t) render them more likely to pass unnoticed in speech. Use of [?] for intervocalic (t) also involves a phonetic difference and not a phonemic merger, yet this feature is highly salient and well above the level of awareness, as evidenced by the extent to which it is commented on in the data. Despite the overt stigmatisation it appears to carry, use of the glottal stop as a realisation of word-medial (t) is a feature which is well established in MbE, such that it is used near categorically amongst young speakers in the sample, as detailed in section 4.2.6. As such, its use is somewhat different from the incidence of [u] for (r), [f] and [v] for (th) and (dh), and [ht] for pre-pausal and turn-final (t) in the data, which are all in evidence, to a relatively small degree, in the speech of the young in the sample only. 236

Speakers’ awareness of changes in MbE offers insight into the evaluation of the forms of interest. As noted in section 5.1.2, in response to the question of whether age-correlated variation is perceived in MbE, none of the 32 informants mentioned [u], despite its widespread use in the sample (recall that 14 of the 16 young informants revealed some use of [u]). This is also despite the fact that intrusive /r/ is noticed and classed as a ‘southern thing’ (intrusive /r/ was found to be overwhelmingly avoided in conversational style in Newcastle (Foulkes 1997: 84)). Likewise, pre-aspirated pre-pausal (t) is not mentioned, although this feature has a restricted use in the sample. However, use of the fronted variants of (th)/(dh) was commented on by two female informants from the old group, despite also having a very restricted use in the sample. As use of the fronted variants of (th)/(dh) was declared by these two informants to be a feature which would be corrected if heard, attitudinal information would suggest that [f] and [v] are indeed more salient in speech and more overtly stigmatised than are [u] and

[ht]. As such, use of the fronted variants of (th) and (dh) is likely to carry covert prestige.

By contrast, the more widely used [u] appears not to carry any covert prestige. If anything, [u] is looked upon with derision. This is demonstrated by the fact that during an interview, two young males, both revealing some use of [u], discussed an acquaintance who ‘couldn’t say her s’, with open scorn expressed for this perceived defect. This suggests that, although they use it, [u] is evaluated negatively when noticed by speakers who value the covert prestige of non­ standard forms. Paradoxically, then, although [u] is the form which is associated with affected speech (in an ‘upper class’ sense), as noted in section 4.3.1, speakers appear unaware of their adoption of it.

It seems that the degree of salience and the covert prestige carried by a variant are not necessarily predictors of which innovatory forms will be diffused into a variety. Forms with seemingly low salience and no covert prestige, such as [u], can be diffused as quickly as a feature with high salience, such as TH-fronting. Although evidence from the sample suggests that the spreading features, [u] for 237

(r) and [f], [v] for (th) and (dh) respectively, appear to have emerged in MbE at the same time and at the same overall level of usage, the spreading features

appear not to be working on the same level, [u] has been widely adopted by the

young speakers in the sample and its use has increased in the speech of the adolescents compared with that of the young adults. It appears not to be noticed in the speech community and seems not to carry covert prestige. By contrast, use of the fronted variants of (th) and (dh) has a very restricted use in the sample and there is no increased use noted between the adolescents and the young adults. The feature is noticed in the speech community and appears likely to carry covert prestige. This may indicate that the change in progress in (r) is operating on a different level of awareness than that which appears to be motivating the change in (th)/(dh) and also that in pre-pausal and turn-final (t).

The informants’ awareness of the form their accommodatory behaviour takes is also of interest in regard to their evaluation of variants. Many speakers who claimed to be aware of shifting to a less localised style of speech in certain circumstances expressed this shift in terms of an avoidance of [?] for (t) and of the zero realisation of (h), both of which are considered to be stigmatised. Interestingly, the possibility of using a closing diphthongal realisation for the

GOAT vowel was rejected consistently, as noted in section 5.1.3. If the southern mainstream diphthong is taken as a prestige variant, this suggests that the speakers do not operate on a vemacular-to-standard continuum. Rather, in shifting to a less localised speech style they simply avoid forms which they perceive as overtly stigmatised. The shift does not compromise the speaker’s identity as a ‘Northerner’, as it appears not to entail use of forms which may be perceived as ‘southern’. The ‘Southern’ closing diphthong does not receive a positive evaluation. In fact, contemplation of its use provokes an amount of derision and antipathy. This would suggest that speakers do not consider a monophthongal ‘northern’ realisation of the GOAT vowel as stigmatised, nor do they perceive themselves to change it in a more overtly careful speech style.

Instantiations of second order indexicality are also evidenced in style shifting, as noted. The linguistic data reveal that informants style shift between different 238 realisations of the variables under investigation in conversational style and word- list style (see Chapter 4 for details). This style shifting can be seen as an ideologically motivated manifestation of linguistic insecurity (see Labov 1972). In shifting from one variant to another in less to more formal speech styles, the informants demonstrate their evaluation of the variants under investigation. Such style shifting indicates that the variant is a marker of linguistic variation. Avoidance of glottal stop realisations of pre-pausal (t) and word medial (p t k) in word-list style reinforces the assessment of overt stigmatisation of the glottal stop, as evidenced in informants’ opinions on age-correlated differences in MbE and their discussion of it as a form they would avoid in more formal contexts. Variants of (r) appear least subject to style shifting in the data, which also parallels the fact that the relatively widespread use of [u] in the speech of the young appears unnoticed in the speech community.

Furthermore, an interesting finding in the data is that the two most innovatory speaker groups of the sample, the young adult males and the adolescent females, demonstrate considerably less style shifting than other groups. In word-list style, the young adult males reveal the highest use of [?] for pre-pausal and word medial (t), categorical use of the localised [?p], by far the highest use of [?k], the highest use of [u], and the only use of the fronted variants of (th) and (dh) of all speaker groups. The only other speaker group which demonstrates similar linguistic behaviour is the adolescent female group. Little evidence of style shifting with regard to use of pre-aspirated pre-pausal (t) is found in the speech of the adolescent females. Additionally, they reveal the second highest use of [?p] in word-list style (all other male groups demonstrated a higher use of [?p] in conversational style than the adolescent females), the second highest use of [?] and the only female use of [?t], the second highest and only female use of [?k], and the highest female use of [u] in word-list style. The speakers of these two innovatory groups do style shift to a degree. However, a tension is revealed between the value placed on the covert prestige certain forms carry and the overt prestige of the forms used by other speaker groups in word-list style. On balance, it appears that the covert prestige carried by the innovatory or increasing forms 239 overrides the tendency to shift to the less localised or stigmatised variant in word-list style. This is exemplified in the word-list reading of one of the young adult male speakers who, when reading the citation form happen, began by producing a [h] word initially, then checked himself and re-read the citation form with a zero realisation of (h). The control that speakers have over the forms used in word-list style is evident in the fact that a number of them comment on the fact that they have altered their speech when they begin to read the citation forms. Consequently, evaluating a variant as a marker of variation due to its being subject to style shifting, as discussed in section 2.3.2, is not unproblematic as, for the data presented here at least, the speakers who are responsible for the adoption of innovatory forms are those who appear least affected by linguistic insecurity as evidenced in style shifting behaviour.

6.2 Linguistic variation as a stylistic resource

The notion of linguistic markets which are part of a broader symbolic market including patterns of consumption and demeanour has been discussed by Eckert (2000). Alternative linguistic markets can be seen to exist in which ‘forms other than the global standard constitute the norm’ (Eckert 2000: 18). In addition to the local orientation and the ideologies underpinning the choices of variants used, the variation in the data, particularly the fine grained age-differentiations in evidence in the four young speaker groups of the sample, leads us to regard linguistic variation as a stylistic resource which is exploited by speakers to indicate their alignment with certain groups within the speech community. None of the changes in progress in MbE suggested by the increased use or adoption of variants by young speakers can be said to represent an orientation towards a prestige variety or standard. As such there appears to be an alternative symbolic youth market which stands in opposition to the ‘standard’ market. Again, speakers’ attitudes and ethnographic profiles are beneficial to an understanding of the constructions of identities through use of linguistic resources available. The fine-grained age differences evident in the data are discussed firstly in section 6.2.1. We then consider gender differences in the data in section 6.2.2, before examining the results of particular innovatory speaker groups in 6.2.3. 240

6.2.1 Age

In section 2.1.3 the use and importance of age as an extra-linguistic variable in apparent time studies of language variation was outlined. It was also noted that speakers are often grouped chronologically into fairly broad age cohorts and assumptions are made about the behaviour of the members of an age cohort corresponding to labels such as adolescence, middle age, old age, etc. However, the broad life stages that these assumptions appear to be based on are fluid, in a sense. Individuals have different experiences of the movement through life stages which gives age meaning. Thus, grouping speakers whose chronological ages fall within a specified age-range does not guarantee that the speakers share the same life stage or that they will conform to assumptions associated with that life stage. Furthermore, these life stages are not fixed entities, but change over time. Speakers may experience life stages differently from those who passed through them a generation ago.

Importantly, the social psychological experience of the ageing process appears to be changing rapidly in Britain. Both internal and external factors can be invoked to account for these changes. Biological factors such as the earlier onset of puberty and better health in older age are clearly contributory factors. However, consumer-oriented factors must also be borne in mind, as they are likely strongly to influence the social psychological experience of the ageing process. For example, the most sought-after consumer in Britain is now the ‘tweenie’, with the pre-adolescent or ‘tweenie’ market estimated to be worth in excess of £30bn (The Observer Dec. 3, 2000). With designer wear, cosmetics and magazines aimed specifically at 8-14 year olds, preoccupation with image is likely to begin at an earlier age. It is, however, equally likely to continue for a longer period, as the ‘middle youthers’ also constitute a huge consumer group in Britain. This is evidenced in the fact that certain British stores now carry ranges designed to allow the ‘middle youther’ mother and her ‘tweenie’ daughter to dress alike. Lifestyle changes appear to be affecting all life stages, and a recent subheading of an article in a British Sunday newspaper on the changing ageing process expressed the situation thus: 241

[w]hile tweenies flex financial muscle, party animals are becoming pensionable and students network in the City rather than going on demos. Age ain’t what it used to be (The Sunday Times, Jan 2001).

The changing social psychological experience of ageing is likely to have an effect on linguistic behaviour. If the ‘middle youther’ mother wishes to dress like her ‘tweenie’ daughter, it is possible that she may also wish to speak like her. Equally, the 21 year old student who networks in the City and worries about job security may be more responsive to the norms of linguistic behaviour and be more concerned with notions of linguistic correctness than the middle-aged speaker who now forms part of the ‘middle-youther’ market. Any effects of these apparent changes may not be immediately perceptible in linguistic behaviour. However, the traditional assumptions of the social psychological effects of ageing that are used to explain age-correlated variation in linguistic behaviour should perhaps not be automatically taken as givens.

To gain further insight into the effects of ageing on linguistic behaviour the speakers of the present study were grouped into four emically-defined cohorts, the 16 young speakers of the sample being separated into two independent age groups representing the two life stages of young adulthood and adolescence, as detailed in section 3.1.3. This separation was motivated by the desire to seek evidence for fine-grained age differentiations in linguistic behaviour between the two life stages and to test the hypothesis that adolescence is the focal point of linguistic innovation and change. As noted in section 2.1.3, adolescence is considered a crucial age for linguistic change. Although considerable evidence is available to corroborate adolescents’ innovative behaviour, many studies reach this conclusion by analysing the speech of adolescents only, or by comparing the speech of adolescents with those of much older speakers (for example Eckert 2000, Cheshire, Gillett, Kerswill and Williams 1999). This will doubtless reveal differences. However, it does not indubitably confirm the singular role of adolescents in language innovation and change. If many of the speaker-based changes in BrE are contact-induced, a problem arises in how forms are diffused through speakers who experience relatively little mobility. As Britain (2000: 57) argues, ‘[t]he existence of a motorway from A to B may help relatively little in the diffusion of a change if the diffusers can’t drive’. As noted also in section 242

2.1.3, young adulthood is considered a time when, under marketplace pressures, a move towards a higher level of use of high prestige variants is observed. Whether such fine-grained age-correlated variation could be revealed was therefore central to the design of the fieldwork sample.

Variation between the young adults’ and the adolescents’ linguistic behaviour was indeed evident in the data. Marked differences between the linguistic behaviour of the adolescents and the young adults were found to be in evidence in most variables. Of interest particularly is the distribution of pre-aspirated pre- pausal (t) and the fronted variants of (th) and (dh) in the data. For both variables the generational differences suggest that the forms, [ht] and [f, v] have emerged in MbE very recently, as their use is confined to the young of the sample. Nevertheless, very restricted uses of the innovatory forms are revealed in speech of the young. Both innovations appear to have been adopted by very specific speaker groups. [ht] as a realisation of pre-pausal and turn-final (t) is used overwhelmingly by the adolescent females, [f] and [v] as realisations of (th) and (dh) respectively are to be found almost entirely in the speech of the young adult males. These speaker groups’ use of the innovatory forms is not only notably different from the use of the forms by the opposite gender, but is also markedly different from that of speakers of the same gender and the same socio-economic class who are a mere three or four years older or younger than them. These findings raise questions about the processes of linguistic change and the role of speaker age and gender in the adoption of innovations.

Considering speaker age, although it is adolescent speakers of the sample who are adopting the innovatory pre-aspirated (t), it is clearly not always adolescent speakers who are responsible for the introduction and adoption of new forms into a variety. The adoption of the fronted variants of (th) and (dh) by the young adult males, rather than the adolescents, to some extent refutes the widely held assumption that adolescence is the focal point of linguistic innovation and change. Furthermore, use of the stigmatised fronted variants of (th) and (dh) is clearly not an example of an increase in standardisation in young adulthood. 243

In the adoption of the innovatory forms [ht] and [f, v], the important finding appears to concern neither the age of the informants nor their gender. Rather the interaction of these two extra-linguistic variables appears to be crucial in the identification of the speakers of the sample responsible for the adoption of innovatory forms. The importance of the interaction of extra-linguistic variables is not a new finding. Indeed, we are often confronted with compelling evidence which suggests that innovations are found in the speech of young females, or in the speech of MC females, and the use of localised forms is associated with old males or with WC males. However, to find such sharply differentiated variation in a group of speakers of the same socio-economic class but with an age range of six years is puzzling and raises issues of interpretation.

Again, speakers’ perceptions and the local knowledge they operate with are useful in this regard. The speakers’ perceptions of age-correlated variation in the speech community, as well as providing insight into the extent of the awareness of linguistic change in progress as noted in the previous section, also denote perceived age differences as opposed to chronological age differences. An observation of particular relevance, noted in section 5.1.2, is that of two of the young adult females of the sample whose immediate response to the question was to stress the perceived difference between themselves and adolescent speakers (both informants were 21). Furthermore, the speech of teenagers was considered to be more non-standard or localised than the speech of young adults. The fact that the young adult females immediately classed themselves as the older speakers and contrasted themselves with speakers who are only a few years younger than themselves is revealing. The distance between perceived age difference and actual, chronological age difference is clearly not always the same, and to consider age differences in terms of chronological age only can be misleading. The fact that the age range of the young speakers of the sample is narrow, recall that we are dealing with an age cohort which spans six years only, emphasises this point. The perceived distance between the ages of 21 and 16 is considerably greater than that between 31 and 26, and 41 and 36 and so on. This is largely because the life landmarks which punctuate the period between age 16 and 21 are likely to be considerable. As noted in section 3.1.3, the emically- 244 defined age cohorts of the study centre upon the fact that, in Britain, the age of 18 represents a rite of passage which is probably unequalled in an individual’s lifetime. The difference between ages 16 and 21 covers the crucial transition from dependence to independence. It seems, then, that fine-grained differences in linguistic behaviour are more likely if the perceived age difference is great, and indeed, the perceived age difference is considerable at this crucial transition period of life.

It is evident that in apparent time studies which seek to find evidence suggestive of linguistic change in progress, grouping speakers through use of chronological age into fairly broad age cohorts can mask specific group effects. What looks like a general trend can be more specifically located, and fine-grained age differentiations can be uncovered in speech. In the linguistic behaviour of speakers defined as young this can prove invaluable in identifying those responsible for the early adoption of innovations. Furthermore, contrary to conventional wisdom, such speakers are not always adolescents. Additionally, analysis of perceived age differences can reveal how speakers group themselves, which, importantly, may not coincide with how they are grouped by the investigator.

Furthermore, the fine-grained age differentiations in evidence in the study imply that speakers use innovatory forms as a stylistic resource, systematically preferring or dispreferring variants to realise their sociolinguistic identity within the speech community. Therefore, whether use of innovatory forms reflects the early stages of change in progress in MbE or cases of age-graded behaviour is not beyond doubt. It is hoped that further light will be shed on this matter through the findings of the panel study currently being undertaken.

6.2.2 Gender

As noted in Chapter 2, the complexity of language and gender is considerably oversimplified by viewing males and females in a dichotomised manner, asserting that males speak one way and females speak another. It was argued that such a view ignores the homogeneity between male and female speech as well as 245 the heterogeneity within male speech or female speech and the interaction between extra-linguistic variables and gender. In the linguistic data we find that the interaction of gender and age does indeed result in the heterogeneity of linguistic behaviour in a single-gender group. This is evident in both the single­ gender generational differences uncovered, which are suggestive of change in progress, and in the fine-grained differences which are revealed between young adults and adolescents of the same gender.

In some cases the variation in the data, both generational and fine-grained, suggests that males and females use resources available to construct sharply differentiated genderlects. This is true both in use of localised and non-localised forms (note the gender differences amongst the older speakers in the variants of (p)) and in use of innovatory forms (note female use of [ht] and male use of [f] and [v]). However, this is not always the case. For some variables, the age- correlated variation in the data suggests convergence of the genders amongst the young speakers. In such cases, the differences between young females and old females, for example, is more marked than the difference between young females and young males. In use of the glottal stop realisations of word-medial (t) we observe the eradication of the marked gender difference in the distribution of variants found in the speech of the older groups. In the speech of the young, the genders appear to have converged on the use of one variant, the glottal stop. Similarly, in the distribution of variants of (r), differences between the young males and young females appear less marked than those between the older males and older females. Additionally, the increase in word-medial glottalisation of (p) by the young females, as well as indexing a North Eastern or specifically Middlesbrough identity, also brings young female speech patterns closer to male speech patterns and differentiates their speech from that of the older females. This may, in fact, be a contributory factor in the motivation for the increased use of localised or stigmatised variants. Shifting gender roles in society have led to a blurring of traditional gendered behaviour patterns. This is amply demonstrated by a consideration of current female role models and high profile media figures. We see numerous female figures, role models and icons, both real and fictitious, who project an active, often aggressive image in the media; ex-Spice Girls, Sarah 246

Cox, Lara Croft, Zoe Ball, Madonna. Indeed, the term ladette has come to be associated with females who exhibit behaviour traits traditionally associated with males, further emphasising the blurring or convergence of gendered behaviour (The Sunday Times 9 August 1998).

Awareness of changing gender roles is indicated in informants’ perceptions of language and gender. However, the unprompted focus of the informants appears to be on female speech patterns and the deviation of female speech patterns from those traditionally associated with their gender. This perceived deviation leads to young female speech being viewed as marked. We noted in section 5.1.2 that the attitude towards female speech has shifted from its being perceived as less localised than male speech according to old speakers, to being as localised as male speech according to middle speakers, to its being negatively evaluated and perceived as more localised than male speech amongst the young male speakers. However, the young female speakers demonstrate a considerably lower level of use of the localised glottalised word-medial (p) than do the young males of the sample. Furthermore, the young males are ahead of the young females in the use of [?] for (t), [f] and [v] for (th) and (dh) respectively, and [u] for (r). Freed

(1996: 70) argues that when the language of male and female speakers fails to meet the expectations commonly associated with their sex, a set of judgements is made about them. Their language is seen as 'marked' and they themselves are viewed as 'deviant.' This is illustrated in the negative evaluation of young female speech patterns and the frequent descriptions of young females as ‘aggressive’ and ‘common’ noted in Chapter 5.

Although the level of use of localised forms by young females is lower than that of the young males, the age-correlated variation in the perceptions of female speech parallels an increased female use of localised or stigmatised forms. What is also of interest, however, is that as we see increases in young female speakers’ use of localised or stigmatised forms, we also note that the young females, particularly the young adult females, are the speakers of the sample who dislike their accent, claim a perceived inability to recognise their accent, and display the lowest levels of local orientation. This may reflect the sensitivity of the young 247 adult females to the consequences of the negative evaluation of failure to meet the expectations of female speech patterns, something which is yet to be experienced by the adolescent females. (Recall the young adult female concern with the perceived adverse affect of a localised accent on gaining employment, noted in Chapter 5.) A tension, then, is revealed between the wish to use language to differentiate themselves from older speakers and the desire to project the right image. To the young adult females this image appears not to be one which is too locally defined, as this deviates from traditional expectations associated with female speech patterns and can, and indeed does, lead to their being perceived as deviant.

6.2.3 Innovatory speaker groups and individual speakers

The fine-grained differentiations in evidence amongst the young speakers of the sample suggest that speakers are able to use linguistic variation as a stylistic resource to identify themselves as part of locally relevant groups. However, in terms of use of innovatory forms, two speaker groups were conspicuous for their enthusiastic adoption of certain variants. These groups, as noted, were the young adult males and the adolescent females. We turn finally to a more detailed profile of the speakers of the sample who appear to act as innovators.

If we consider the young adult male use of the fronted variants of (th) and (dh) firstly, the dissimilarity in the use of [f] and [v] between the two groups of young males is unusual, particularly given that the adolescent males demonstrate the highest use of [u] of any group. Recall, however, that [u] passes largely unnoticed in speech and appears unlikely to carry covert prestige.

The importance of mobility and face-to-face contact between speakers of different varieties in language change has been highlighted in section 2.2.1 (Trudgill 1986). The fact that the young adult males are more geographically mobile than the other young speakers is likely to be a contributory factor in their adoption of the innovatory forms. In the present economic climate, Middlesbrough has more than twice the national average unemployment rate, as 248 noted in section 3.1.1.5. As such, the young often leave in search of employment, which frequently comes in the form of short-term contracts, thus necessitating their return to the area. Indeed, in a recent report by the Department of the Environment (1996), Middlesbrough was found to be an urban centre with a high population turnover in which young adults are considerably more geographically mobile than average. Despite the town being somewhat isolated, then, its inhabitants often experience appreciable geographical mobility, and thus prolonged exposure to other varieties of BrE. Three of the four young adult males of the sample have lived and worked outside the area in recent years (see Appendix 8 for details). This will have brought them into increased contact with other varieties of BrE. As such, they could be seen as ‘language missionaries’ or weak ties in the community through which innovations can filter, as discussed in section 2.2.1.

Another explanatory factor could lie in the fact that the young adult males of the sample seem exceptionally image-conscious and are perceived by young females to be more image-conscious than females, as discussed in section 5.2.3. This male preoccupation with image in Middlesbrough is to some extent reflected in the retail make-up of the town. It is perhaps no coincidence that a somewhat depressed, medium-sized town has retail outlets which win national designer retailer awards and are visited by internationally renowned designers, 70% of whose business is in menswear (manager of ‘Psyche’ designer clothing store, personal communication). This preoccupation with image might extend into language use. The covert prestige carried by TH-fronting might make it attractive as an urban way of speaking, which arguably has no regional associations for the speakers, either locally or nationally. Rather, the fronted variants of (th) and (dh) seem to be associated with youth culture, the extent of which is demonstrated by phrases used in the media such as YOOF T.V. iyoof representing the word ‘youth’ with the fronted variant of /0/). It may be that use of [f] and [v], with the covert prestige it carries, appears attractive to those most preoccupied with image, the young adult males. It does not, however, compromise their identities as ‘North Eastern’ for, if it did, it would presumably be rejected by these 249 speakers, demonstrating, as they do, the highest levels of local orientation and affiliation of the whole sample. Explanations for the dissimilarity between the adolescent females and young adult females are considerably more opaque. If the adoption of pre-aspirated (t) is a contact-induced phenomenon, it is not immediately obvious why the adolescent females would have more contact with speakers with pre-aspiration, presumably speakers from further north, than the young adult females. However, perceptions of variation are again interesting in respect of the dissimilarity between the two groups of young females. In the linguistic data, we observed that the adolescent females made considerably greater use of glottalised word- medial stops than did the young adult females. They also revealed a near categorical use of [?] for word-medial (t). Interestingly, young speakers of the sample show some awareness of this fine-grained age-correlated variation. (Recall the two young adult females who immediately contrasted themselves with adolescent females who were said to be ‘more common’, and the two young adult males who specified an age range of 12 to 18 in which females use a broader accent than males.) It seems that not only are fine-grained differentiations present in the linguistic behaviour of the young speakers, but that speakers are aware of this fine-grained variation.

6.3 Summary

In this chapter we have observed how a deconstruction of identity through analysis of attitudinal information can allow and inform a more insightful interpretation of linguistic trends. Conclusions that are drawn through a correlational analysis alone might fail to discover the complexities of the use of language in the construction of identity. How language is used to establish and maintain community identity and, moreover, to reflect changes in this community identity, has been revealed in the findings. Similarly, we have observed how language variation is used as a stylistic resource to construct cohort identities within the community identity. By analysing the instantiations of second order indexicality in the data, we can reveal what linguistic forms index to the speakers. In examining speakers’ local orientation and local knowledge, we come closer to understanding what symbolic value linguistic forms have for the speaker, and thus closer to understanding the motivation for their adoption or rejection. Chapter Seven Conclusions

This study began with two broad aims. In this chapter we shall assess how far those aims have been fulfilled and what conclusions can be drawn from the results of the research. The first objective of the study, following the brief of the scholarship through which the research was undertaken, was to create an innovative method of data elicitation for use in a large-scale, collaborative survey of language variation in the British Isles. The second aim involved investigating the hitherto unresearched urban variety of Teesside, a region situated on the North East coast of England, to assess how far it has been affected by the current widespread consonantal changes spreading, arguably from a South Eastern epicentre, throughout BrE. Moreover, central to the study was the question of whether adoption of the spreading forms was concurrent with a loss of localised forms, thus leading to homogenisation on a broad scale. Furthermore, which speakers of the sample were responsible for the adoption of innovatory forms and whether their use compromised or affected the speaker’s regional and social identity were principal concerns.

Considering the methodological aim first of all, the study has revealed that the elicitation of comparable data on lexical variation can be included to effect in a technique designed to elicit casual speech in a relatively rapid manner. The method of data elicitation designed and trialled in the present study demonstrates that multi-levelled comparable data can be captured in a relatively expeditious and uncomplicated procedure. That sufficient data are elicited through the method to allow for a phonological analysis of speech classed as casual style is amply demonstrated by the findings of the study. Although lexical variation is not a focus of the research, the three completed SRNs presented as Appendix 3 demonstrate the quantity of lexical data procured through the method. Use of the innovative SRNs allows for the acquisition of the written record of the regional and social lexical variants (for use in the study of non-standard orthography), the 252 recorded spoken record of the lexical variants (for use in the phonetic transcription of the lexical variants), and the discussion of the lexical variants in a casual speech style (for use in phonological and grammatical analyses of speech classed as conversational style). Moreover, the discussion of the lexical variants produces a mass of attitudinal information on the lexical variants elicited. This can include information on knowledge and use of the variants, connotations and collocations associated with the variants, perceived social variation in use of the variants, and perceived origins and changing societal attitudes towards the lexical items. As such, that which is elicited is not simply comparable lexical variants, but is evidence of how the comparable lexical variants are used.

The second question asked of the methodological aim of the study concerned the possibility of the effective inclusion of attitudinal information in the already multi-layered data elicitation technique. Such attitudinal information would allow for qualitative analyses of the data which would augment any quantitative analyses. Again, this has been successfully demonstrated, as the findings presented in Chapter 5 illustrate. Attitudinal information can, to some extent, be quantified, revealing socially-conditioned differences in perceptions of variation. Such socially-conditioned attitudinal data can be correlated with linguistic data. The instantiations of second order indexicality, as revealed through overt and covert noticing and discussion of language, allow insight to be gained into the ideologies underpinning the speaker’s choice of linguistic variants. Such ideology, Woolard argues, ‘needs to be analyzed systematically in the study of language, not invoked opportunistically or dismissed summarily’ (1992: 236). Analysis of the speakers’ self-images through definition and delimitation of speech community, perception and awareness of variation within this speech community and degrees of local allegiance and orientation allow for a more insightful interpretation of the motivating social factors underlying speaker-based linguistic variation and change than is achievable through a traditional correlation analysis.

A number of questions were asked of the second aim of the study, the empirical investigation of a previously unresearched urban variety of BrE. Given the 253 widespread consonantal changes spreading throughout urban varieties of BrE, whether any of the forms associated with the South East of England had reached Middlesbrough, lying some 236 miles from London, was investigated. Certain of the forms were indeed found to have made inroads into MbE. Apart from the near categorical use of [?] for word-medial (t) amongst young speakers of the sample, [u] as a realisation of (r), and [f] and [v] as realisations of (th) and (dh) respectively were in evidence in the data. However, whether or not this can be viewed as evidence of the broad-scale homogenisation of BrE is questionable. Evidence was also uncovered for the simultaneous increasing use of localised forms, specifically a globalised realisation of word-medial (p), and the existence of what can be viewed as a regional innovation, [ht] as a realisation of pre-pausal and turn-final (t). Whilst the recent diffusion of innovatory forms into MbE is evident, this does not entail wholesale levelling of locally marked variants in the variety with the consequent reduction of the number of variants of phonological variables available to speakers. Indeed, the variant pool is increasing. A similar situation was found in Corby, where ‘movement towards homogenization or leveling is counterbalanced with a movement towards establishing new local norms’ (Dyer 2000: 163). It is evident that the supposed rampant spread northwards of the consonantal features associated with London and the South East of England, or the media-created ‘beast known as ‘Estuary English” as Britain (2000: 54) puts it, does not automatically entail the loss of locally marked variants in the variety as a whole. Innovations can be adopted and locally marked forms disfavoured in a single variable (note the distribution of variants of (r)). However, this is counterbalanced by an increased use of localised forms of other variables. Innovations can thus be adopted without entailing a broad-scale levelling tendency resulting in widespread homogenisation.

The investigation of MbE also entailed analysis of the nature of socially- conditioned variation in the consonantal variables of the study. The distribution of variants of all variables analysed was revealed systematically to correlate with the social characteristics of the speakers of the sample. Age- and gender- correlated variation was in evidence throughout. The decisions to include four age cohorts in the study at the expense of the extra-linguistic variable of socio­ 254 economic class was wholly justified, given the age-correlated variation uncovered in the data. Generational variation suggestive of change in progress in MbE was uncovered by analysing the speech of the young compared with the middle and old speakers. However, fine-grained age differentiations were also revealed between the two independent age groups of young speakers investigated. Furthermore, it was revealed that the generational trends suggestive of change in progress could be more specifically located, as speakers responsible for the early adoption of innovations could be identified more precisely. Moreover, such speakers were not always adolescents as is commonly postulated. Thus the division of the 16 young speakers of the sample into two cohorts corresponding with the life stages of young adulthood and adolescence proved profitable for testing and partially refuting the hypothesis that adolescence is the focal point of linguistic innovation and change.

Gender-correlated variation was also uncovered in the data. In many cases, this followed the pattern, customarily revealed in variationist studies, of males preferring localised variants and females showing a preference for non-localised forms. However, in some cases, the differences between the young speakers and the older speakers of the same gender were more marked than the differences between the young males and the young females, suggesting a blurring of traditional gender roles and a convergence of gendered behaviour.

Whether socially-conditioned variation correlated with speakers’ identity construction and local orientation was the final question asked of the research. Speakers can be seen as constructing their identity linguistically in terms of the larger community they perceive themselves to be part of, broadly correlating with the notion of the speech community, and in terms of the smaller cohort identities within this speech community. Speakers’ perceptions of self-image and awareness of variation were found not only to correlate with the socially- conditioned linguistic variation revealed in the study, but were also reflective of changes in society as a whole.

The shifting identity of the urban centre under investigation was mirrored in the age-correlated shifting self-images of the informants. Moreover, the changes in 255 reactions to salient groups revealed across age were reflected in the linguistic trends uncovered. An increased use of what can be viewed as a salient form in the North East of England, glottalisation of voiceless stops, was revealed in the speech of the young of the sample. The convergence of MbE with varieties further north that this increased use of glottalisation suggests is synchronous with the increasing association of Middlesbrough with the North East of England and dissociation of it with Yorkshire. However, analysis of speakers’ local orientations within a language ideology framework revealed that this convergence does not entail an act of identity or a positive identification with the dominant urban centre of the North East, Newcastle, which may have been the logical conclusion arrived at without recourse to speakers’ local knowledge and local orientation necessary for a deconstruction of identity.

The ethnographic approach used in the study also allowed insight into the question of whether the speakers’ sense of identity and local orientation are affected by the adoption of ostensibly ‘southern’ features which form part of the current widespread vernacular changes in BrE. The results of one particular group of speakers, the young adult males, were consequential in this regard. The speakers revealed to be virtually wholly responsible for the use of the fronted variants of (th) and (dh) in the data, a feature which was noticed and stigmatised within the community, were the young adult males. These speakers also demonstrated the highest level of use of [?] for word-medial (t) in the data and substantial use of [u]. Furthermore, the young adult males were the speakers who were revealed to have experienced the most extensive and recent geographical mobility, thus bringing them into contact with other varieties of BrE. Nevertheless, these speakers unexpectedly demonstrated the highest levels of local affiliation and local orientation of all speaker groups of the sample. Moreover, the young adult males also exhibited the highest level of use of the localised variants of (p) and (k) in the data. It appears, then, that adoption of forms which are associated with the South of England does not affect or compromise the speaker’s sense of identity as Northern or North Eastern. Neither, arguably, does short-term geographical mobility. On the contrary, a heightened sense of local affiliation and identity may be effected by increased 256 geographical mobility. It can be argued that the significance of the individual’s regional identity is intensified when it becomes counter-norm, that is, when the individual relocates. The increased significance of regional identity to the speakers who have experienced recent geographical mobility could surface linguistically as an increased or exaggerated use of localised forms. Rather than increased mobility leading to sameness and homogenisation, it could lead to increasing heterogeneity as speakers adopt forms which carry symbolic value to them, yet simultaneously increase localised forms as their regional identity becomes more salient and significant to their identity construction.

Chambers argues that ‘[t]he underlying cause of sociolinguistic differences, largely beneath consciousness, is the human instinct to establish and maintain social identity’ (1995: 250). It follows then that explanations of sociolinguistic differences should begin with attempts to deconstruct these social identities. This study has revealed that by attempting to access local knowledge and the symbolic function of linguistic forms, a more insightful interpretation of sociolinguistic differences can be proposed. Perceptions of place, perceptions of age and perceptions of gender have all been found to be germane to an interpretation of the linguistic variation uncovered. Furthermore, social changes both in the identity of the locality under investigation and in terms of the social- psychological experiences of ageing and gendered behaviour play a part in the construction of identity and thus in linguistic variation. Milroy (1992: 28) argues that there is ‘no reason why we should not inquire further into social structures and processes as part of an inquiry into linguistic change, and no reason why we should not also think of linguistic change as being an aspect of social change in general’. Without recourse to the local knowledge and social context in which speakers operate, explanations for the motivating factors underlying speaker- based linguistic variation and change may prove fallacious.

A considerable amount of future work on Teesside English would be relevant and timely, most particularly analysis of the vowel system which would incorporate analysis of internal and external factors in linguistic change. The centrality of language in identity construction and the significance of identity construction to an explanation of linguistic variation have been demonstrated by 257 this study. The investigation of linguistic variation in a locality in which identity construction is complex and fluid is clearly beneficial to the examination of the interdependence of language and identity. Middlesbrough, and indeed the larger conurbation around the River Tees, is such a place.

Middlesbrough has always been and continues to be indisputably a border settlement. Geographically sitting on the northern edge of Yorkshire it remains, in spite of its economic and administrative past as a ‘Teesside’ or ‘Cleveland’ town, a place with a distinctiveness and independence all its own, having successfully resisted all attempts to date to treat it as a county capital, Yorkshire ‘outpost’ or Durham dormitory town (Turner 1996: 100).

The ‘distinctiveness and independence’ of Middlesbrough is not unique. Indeed, all border localities and urban centres with substantial in-migration may be disposed to make the same claim. Such localities offer fertile ground for the study of the role of linguistic variation in identity construction, which may, indeed, be an underlying cause of sociolinguistic differences. 253

Appendix One: Biographical Data sheet

BIOGRAPHICAL INFORMATION

name " S i z n .rxA^.....& a j d d ^ . -

sex ......

age (dob).... ^ .....fc - - ...... place of birth..... iXJX/Cj. l o ...... birth place of mother- .’r i . a F f l . e f ’o c . C ...... birth place of father...... H a ...... birth place of grandmothers...... •HQ£W-6pCO-L'...... W^UfScp.C-O./.... birth place of grandfathers ...... GcX^. U j.

ethnic group...... (^.{nvJCX...... occupation (current / usual).... H_C/~ j L^CXAXtr^C-......

assessment of social class ...\& crfkSjruz .... CJ-CkSS...... b o m - , . . . ? . housing...... 0...kxjr\sj^ ...... education tVyJj. h-il/T. identification score index i A A A - . W ..... i g .... 5 g ...... b c . l f : 259

Appendix Two: Instruction sheet

First Name ...... s i £ n n ^ ......

Place of b irth ... -*• dcl..ljLSh..

Other places you have lived and for how long ^ tS L S djQJjL ...... h .|£ & £ ...

n k^..L^x.v^ ...... ^ ..J A tL a n ..... ( i k l i

• Please complete the sheets with words you think are dialect words or are local to fee area you are from.

Try to put down the first thing that comes to your mind, words you use every day when talking with friends, for example.

• After that, think about it for a while and note down any other examples of words local to the place you live which come to mind.

Feel free to discuss the words with other people from the same area as you. But try to keep a note of who you discuss the words with (especially if you note down their suggestions).

• Put down more than one word, if you like. Also, feel free to use expressions as well as single words.

Use the sections called ‘any others’ to note down any extra words or expressions you think of (yourself, or in discussion with others). If these are words for things not listed on the sheet, please put down what you think they mean, or what someone not necessarily from your area would understand by them.

• Have a look through the questions about your language and your area, which we’ll also be talking about (there is no need to answer these questions on the sheet).

Complete the Language Questionnaire by putting ticks in the appropriate boxes. 260

Appendix Three: Three completed SRNs

Q0>W\ I intelligent^ v stupid '/7(V>WLA soft shoes c t a p f J (worn by children for P.E.) r, rude

n a n ; 4 w S £ ,' 1 f * glasses

W ' if 9Q ~ “ " v y - i r mean unattractive______I 2appearance — tall (with money) p /c u n )^n)CLj

men s facial hair / StlK.G U any'others attractive' (above lip & in PUfcG tlJ la? 5 front o f ears) I.) C(KV\riO^ fee Ax trousers sic k be any others o U s f r d Trado bdtfyLs tkv< 5 ^ ^

PEOPLE nose sruTok I M W partner 3ob (sexual) male / female man / A/Canjj m tu woman head cnxf f o ^ body • ears t i d e - ^ ( 7 **■ fesfe* a Gl /a s s k d . h m t fiu ^ mother / father legs' boss ages O ur -pi*\s & any others ^ I f z r relationships our Dad/oh S fd ld h iixb u baby brother / sister

b c^-bb °ur Kid b c u r r t our loss child (boy/girl) any others

l< v s s ^ ™ 3 r kffforffrr (K ■ In f i grandmother / father o ) h • /?> ru u w cL ( s w i c k a / ) 261

jpllolu ° \ [ ^ |a f ' f(M d - ^ \ d £ r t u pregnant ^OtiCkad mad drunk _ p ^ o ^ J . C s / o zkjzA) SisLMj-td cheated (e.g. financially) tired - Z c W ^ d " ----- d ( M & K K o L ck sM j tx f _ pleased / prolid i»i ± 3 3 H S ! c k a J c h u f f e d fQMlhs)/ h tKsved not have any money left cold . ' ' ->I^UU S kl-yJr brifK C . n^ t n > \ f r o z a * - U_j K U i( j ( U C/&, P (XSfeAj any others - g faczrzJstc( 4 tu U h r OuJUi ^ ul&jl 6 / ^ c u ik u ^ y i s x LINGS, ACTIONS C / a ^ & STATES

a^k to wail h.cuy g (XbouJ~ KftAO^ O v talk / chat run away from (escape) (a lot) . tell on /u cJc (P&- som eone Sccf^-pjy aoM> u (tales) thank 5"pKX-£j ' steal t W

policeman main room of the house kotfaj Q9ff2qT . (with T.V) (b tn c j cUajjuq, r< m K police station

Ccrf> S h&f?

prison small walkway (path) between houses d c h . |C *v\.ck. any others

b *

3 ulK \^ tk

(heavy) W rK u'uu affUM p u t t e r c t vi^k food taken to dcfl/jv work kixM t> o >

meals of dog the day j P o o l L time between (very) i l H M T summer & small stream winter i e d k S i k t f c a Z u p f o r M c k ^vcl any others,h types money (in general) -fde.et' d ftW s-IGue -5>^. 3 w f pasU ZqQ$(MtmjL4) bi<^\o-c,s StrpL ('lilU 5^% ) 263

Appendix Four: Identity Questionnaire

Your Language

• What accent would you say you had, and do you like it? • Can you recognise the accent of Middlesbrough (e.g. if heard on the radio or TV)? If so, how? • Do you think older and younger people talk the same here (pronounce things the same and use the same words)? • Have you ever been in a situation where you’ve deliberately changed the way you talk? If so, why? • Do you think there’s a difference between how males and females speak here? • Where, geographically, would you say people stop talking the same as you and start sounding different? • What would you think if your accent was referred to as Geordie or Yorkshire?

Your Area

• If you were watching a regional news programme, what places would you expect to hear news from? • Do you remember when the county of Teesside was formed and Middlesbrough was no longer in Yorkshire? Do you think this change made a difference? • Would you consider Teesside to be in a larger ‘north-eastern’ part of the country or a larger ‘Yorkshire’ part of the country? Why? • What image or description of Middlesbrough would you give to someone who didn’t know it? • If you wanted a day out shopping, where would you go? • Do you think Middlesbrough is a fashionable place to be? • What do you consider the local football derby to be? • If you could, would you change where you came from? Why/why not? • What do you consider the best and worst things are about growing up and living in Middlesbrough? • Have you ever seen Middlesbrough on a national T.V. programme (e.g. a documentary)? If so, how was it portrayed? • If an outsider was complaining about Middlesbrough, would you defend it even if you agreed with what s/he was saying? Why/why not? • How many friends, relations and work/school/college mates do you have in the neighbourhood (not more than about 5 mins. away) who you see regularly? 264

Appendix Five: Completed Language Questionnaire

Language Questionnaire

2. 0 □ □ They can’t do nothing without you saying. 3- jzf'n □ There’s a job going at our place if youse two want to go for it. z f u □ We all talk different. You weren’t stood there, were you"7 zfn n □ □ □ Just say what you want, innit? 7- tzT□ □ They said they were coming back on Monday and they never 8- □ □ □ That’s the best one what she’s got on. 9. □ □ You’re insured on them items for 80 days 10 □ □ □ He’s working 9 while 6 this week. n □ I’m going down London next week 1 2 ^ 0 □ 1 don’t fancy going up Stockton. is. □ □ D The sharks were only two foot long. 14V° □ □ I seen Sarah at work yesterday. 15. □ □ □ 1 knew a bloke who were doing speech therapy. 16. □ □ □ We was walking along the road when it happened. It were too cold to go out. 17 3 3 n 1 8 ,'O J^ D We usually gan down the pub on Thursday’s. 1 9 .^ * 0 □ I bet she was sick as. 20 «ET □ □ They give me it the same day I opened the account.

21.vcfa □ I should’ve went to the medical really. 22. □ □ D You wasn’t listening to what 1 said. 23. □ □ D She come in at 12 o’clock last night. 24 □ □ □ She don’t like that sort of tiling. 25. □ □ □ There’s no Electron signs on any doors. 26. □ □ □ I’m not cooking for diem, they can do it dieirselves. ?7 jrTy^‘ □ Lend us your catalogue, I want to have a flick through it 28.»Zf □ □ There was kids there. 29 '□ □ I’ve never heard of him like. 30 >jzrn □ He said it wasn’t scary but, mind you, he is about 45.

31 j z f □ □ They proper hurt you when you crash. 32. □ □ □ The cops ain’t gonna do anything. 33. □ □ □ They in’t gonna pull you up. 3 4 v Z fn □ It’s the only like decent night out we have, isn’t it? 35. □ □ □ He wouldn’t could’ve worked, even if you had asked him. 36- □ □ □ Will I put the kettle on? 37 ■ □ □ □ My hair needs washed. 38- □ □ □ I’m opening another account me. If you’re left-handed, you’re more cleverer. I’ve forgot my money, can you buy me a pint. vJZl □ □ 265

Appendix Six: Affiliation Score Index

1. If you were on holiday and saw someone you had never seen before but thought they came from your home town (e.g. you overheard their accent and recognised it, they were wearing the local football shirt etc.), would you: a) feel compelled to go and ask where they were from and strike up a relationship (3) b) feel you had something in common but not do anything about it (2) c) not feel any differently than you would towards any other stranger (1)

2. Would you say you feel close to and feel you have something in common with people from your home town in general (that is people you don’t know personally), or would you say you do not feel any closer to them than to people from somewhere else? a) feel closer to people from home town (3) b) don’t feel any closer to people from home town than to other people 0 ) c) don’t know, can’t say (2)

3. If you were the manager of a company which was recruiting people and two equally qualified and experienced people applied for the position, but one had been bom and educated in your home town and the other had been born and educated somewhere else, would you choose: a) the person from your home town (3) b) the person from somewhere else (1) c) don’t know, wouldn’t matter (2)

4. Would you prefer your child’s school teacher to be: a) a local person with a local accent (3) b) a person who spoke ‘standard’ English with a ‘standard’ accent (1) c) it wouldn’t matter what accent they had (2)

5. If you were voting in a local election, would the fact that a candidate was a local person persuade you to vote for them? a) yes it would (3) b) no it wouldn’t (1) c) don’t know (2)

6. If you wanted to leave something to a charitable organisation would you choose: a) a local one (3) b) a national/international one (1) c) don’t know, depends on the cause (2)

7. If there was a programme on T.V. about your home town which clashed with your favourite programme and you couldn’t record either, would you: a) watch it and miss your favourite programme (3) b) watch your favourite programme and miss the other (but wish you hadn’t) (2) c) watch you favourite programme and miss the other (but not mind) (1) 266

Appendix Seven: Word List

gate seethe friday paint read through later reap whole butter reek photo bottle marry lead tickle carry sheet paper sorry peat happen dream house wicked I hit it happy automatic I pack it hill doctor I sack it apprehend whisper I lap it pillow whisker I tap it milk jumper I beat it fall bunker soap tall I bought it goat I caught it home I wrote it motor I got it purse think worse thing nursing pathetic cursing weather face everything time both pint though knife the ninety then dark mother shark father parking either marking breathe law and order 267

Appendix Eight: Localness charts

Old female Ethel Mildred Hilda Doris

Birthplace Informant M’bro M’bro M’bro M’bro Mother M’bro M’bro St. Andrews Selly Oak Father Glasgow South Bank M’bro Not known Grandmother M’bro M’bro Italy Not known Grandfather M’bro M’bro Italy Not known Grandmother Glasgow South Bank Ayrshire Not known Grandfather Glasgow South Bank M’bro Not known

Mobility Duration -- - - Place -- - -

Old Male Arthur Tommy Keith George

Birthplace Informant M’bro M’bro M’bro M’bro Mother M’bro Sheffield Gramlington M’bro Father M’bro Bradford M’bro M’bro Grandmother M’bro Sheffield Gramlington Scotland Grandfather M’bro Sheffield Gramlington Scotland Grandmother M’bro Bradford M’bro Not known Grandfather M’bro Bradford M’bro Not known

Mobility Duration - 2 years 2 years 2 years Place - Rosedale In army? Japan & Luton 268

Middle female Sally Christine Jill Pauline

Birthplace Informant M’bro M’bro M’bro M’bro Mother Blackpool M’bro M’bro M’bro Father Stockton M’bro M’bro M’bro Grandmother St. Helen’s Essex M’bro M’bro Grandfather Lancashire Ireland M’bro M’bro Grandmother Stockton M’bro Not known M’bro Grandfather Stockton M’bro Not known M’bro

Mobility Duration 2 years 1 year - - Places Brighton Newcastle - -

Middle male Mick Steve A Steve B Barry

Birthplace Informant M’bro M’bro M’bro M’bro Mother M’bro M’bro M’bro M’bro Father M’bro M’bro M’bro M’bro Grandmother M’bro M’bro M’bro Newcastle Grandfather M’bro M’bro M’bro Newcastle Grandmother M’bro M’bro M’bro M’bro Grandfather M’bro Glasgow M’bro M’bro

Mobility Duration 2 years -- - Places Surrey - - - 269

Young Adult Beth NicA NicB Anna Female

Birthplace Informant M’bro M’bro M’bro M’bro Mother M’bro M’bro M’bro M’bro Father M’bro M’bro M’bro M’bro Grandmother M’bro M’bro M’bro M’bro Grandfather M’bro M’bro M’bro M’bro Grandmother Manchester M’bro M’bro M’bro Grandfather Liverpool M’bro M’bro M’bro

Mobility Duration - - 2 years* - Places - - Milton Keynes - * The 2 years in Milton Keynes were experienced when the informant was a small child

Young Adult Paul Pete Kevin Raj Male

Birthplace Informant M’bro M’bro M’bro Pakistan Mother M’bro M’bro M’bro Pakistan Father Spain M’bro M’bro Pakistan Grandmother Scotland M’bro Ireland Pakistan Grandfather M’bro M’bro M’bro Pakistan Grandmother Spain M’bro Ireland Pakistan Grandfather Spain M’bro M’bro Pakistan

Mobility Duration 2 years 1 year 2 years - Places Luton Newcastle South Shields & " Germany All the mobility recorded by the young adult males is recent, it being experienced since leaving school. 270

Adolescent Debbie Lauren Jo Vaj Female

Birthplace Informant M’bro M’bro Halifax M’bro Mother M’bro Burnley M’bro Scarborough Father Coventry Redcar M’bro M’bro Grandmother Norway West Drayton M’bro Scarborough Grandfather Norway Burnley M’bro Scarborough Grandmother Coventry M’bro M’bro M’bro Grandfather Coventry M’bro M’bro M’bro

Mobility Duration - - - - Places -- - -

Adolescent Jonny Dave A Dave B Carl Male

Birthplace Informant M’bro M’bro M’bro M’bro Mother M’bro M’bro M’bro M’bro Father Spain M’bro M’bro M’bro Grandmother Scotland Not known M’bro M’bro Grandfather M’bro Yorkshire M’bro M’bro Grandmother Spain Not known M’bro Durham Grandfather Spain Yorkshire Kings Lynn M’bro

Mobility Duration - --- Places - --- 271

Appendix Nine: variants of (th) and (dh) other than [0], [f] and [8], [v]

Variants of (th) other than [0] and [f], (total number of tokens recorded per speaker group)

Om O f Mm M f YAm Y A f Adm A df 0 1 1 4 6 1 4 0 4

[6] 1 5 5 3 0 4 3 2 m 1 1 0 1 1 0 0 1 [n] 0 1 0 0 0 0 0 0 [?B] 1 0 0 0 0 0 0 0

[t] 1 0 0 0 0 0 0 0

Variants of (5) other than [5] and [v], (total number of tokens recorded per speaker group)

Om Of Mm Mf YAm Y A f Adm A df

0 5 0 3 2 6 7 9 9

[0] 0 1 0 4 2 0 0 1 [d] 1 0 0 0 0 0 0 0 References

Aitchison, J. (1997) The Language Web. The Power and Problem o f Words. Cambridge: Cambridge University Press. Allen, W. (2001) “Totally Mint”: An analysis of (t) in the speech of Tyneside adolescents. Paper presented at The Third UK Language Variation and Change Conference, University of York, July 2001. Altendorf, U. (1999) Es(h)tuary English: Is English going Cockney? Moderna Sprdk. XCIII(l): 1-11. Anderson, B. (2000) Towards an Integrated Model of Internal and External Constraints on Language Change: The Case of the Detroit AAE Vowel System. Manuscript. Anderson, B. & L. Milroy (1999) Southern changes and the Detroit AAVE vowel system, Paper presented at NWA V 28, Toronto. Anderson, G. (1997) The Use and Development of Invariant Tags in London Teenage Speech. Paper presented at The First UK Language Variation Workshop, University of Reading. Andresen, B.S. (1968) Pre-glottalization in English Standard Pronunciation. Oslo: Norwegian Universities Press. Atkinson, Rev. J.C. (1868) A Glossary of the Cleveland Dialect: explanatory, derivative and criticism. London: J.R. Smith. Barltrop, R. & J. Wolveridge (1980) The Muvver Tongue. London: Journeyman Press. Barry, M.B. (1981) Aspects o f English Dialects in Ireland. Volume 1. Papers arising from the Tape-recorded Survey of Hiberno-English Speech. The Institute of Irish Studies, The Queen’s University of Belfast. Beaken, M.A. (1971) A Study of Phonological Development in a Primary School Population in East London. Unpublished PhD. dissertation, University of London. Beal, J. (1993) The Grammar of Tyneside and Northumberland English. In J. Milroy & L. Milroy (eds.) Real English. The Grammar o f English Dialects in the British Isles. Harlow: Longman, ppl87-213. 273

Beal, J. (1999) ‘Geordie Nation’: Attitudes to Dialect and Regional Identity in the North-east of England. Paper presented at the 10th International Conference on Methods in Dialectology, Newfoundland, August 1999. Beal, J. & K. Corrigan (1999) Comparing the present with the past to predict the future for Tyneside (British) English. Newcastle and Durham Working Papers in Linguistics 6: 13-30. Becker, H. (1963) Outsiders. New York: The Free Press. Bell, A.M. (1860) The elocutionary manual. Cited in B.S. Andresen (1968) Pre- glottalization in English Standard Pronunciation. Oslo: Norwegian Universities Press. Bergvall, V. (1999) Towards a comprehensive theory of language and gender. Language in Society 28: 273-93. Bergvall, V., J. Bing & A. Freed (1996) Language and Gender Research: Theory and Method. London: Longman. Bloomfield, L. (1935) Language. London: Allen & Unwin Ltd. Briggs, A. (1996) Middlesbrough: the Growth of a New Community. In A.J. Pollard (ed.) Middlesbrough Town and Community 1830-1950. Stroud: Sutton Publishing Ltd., ppl-31. Britain, D. (1997) Dialect contact and phonological reallocation: ‘Canadian Raising’ in the English Fens. Language in Society 26: 15-46. Britain, D. (2000) The difference that space makes: an evaluation of the application of human geographic thought in sociolinguistic dialectology. Essex research reports in linguistics. Department of Language and Linguistics: University of Essex, pp38-82. Brown, P. & S. Levinson (1979) Social Structure, Groups and Interaction. In K.R. Scherer & H. Giles (eds.) Social Markers in Speech. Cambridge: Cambridge University Press. Butler, J. (1990) Gender Trouble: Feminism and the Subversion o f Gender Identity. New York: Routledge. Cameron, D. (1995) Rethinking language and gender studies: some issues for the 1990s. In S. Mills (ed.) Language and Gender: Interdisciplinary Perspectives. New York: Longman. 274

Cameron, D. (1997) Performing gender identity: Young men’s talk and the construction of heterosexual masculinity. In S. Johnson & U.H. Meinhof (eds.) Language and Masculinity. Oxford: Blackwell, pp47-64. Cameron, D. (1990) The Feminist Critique of Language: a reader. London: Routledge. Chambers, J.K. (1995) Sociolinguistic Theory: Linguistic Variation and its Social Significance. Oxford: Blackwell. Chambers, J.K & P. Trudgill (1980) Dialectology. Cambridge: Cambridge University Press. Chase, M. (1995) The Teesside Irish in the Nineteenth Century. Cleveland History: The Bulletin of the Cleveland and Teesside Local History Society 69: 3-23. Cheshire, J. (1982) Variation in an English dialect: a sociolinguistic study. Cambridge: Cambridge University Press. Cheshire, J., V. Edwards & P. Whittle (1989) Urban British Dialect Grammar: the question of dialect levelling. English World Wide 10(2): 185-225. Cheshire, J., A. Gillett, P. Kerswill, & A. Williams (1999) The Role of Adolescents in Dialect Levelling. Final report submitted to the Economic and Social Research Council, June 1999. Chirrey, D. (1999) Edinburgh: descriptive material. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp223-229. Coates, J. (1986) Women, Men and Language: A sociolinguistic account of sex differences in language. Harlow: London. Coates, J. (1993) Women, Men and Language (2nd edn.). London: Longman. Coates, J. (1996) Women Talk. Oxford: Blackwell. Coggle, P. (1993) Between Cockney and the Queen. The Sunday Times Wordpower Supplement. Costa de Oliveira Filho, M. (1999) An inquiry into the incidence and perceptions o f English TH-Fronting. Unpublished M.A. thesis, University of Leeds. Cohen, A.P. (1985) The Symbolic Construction o f Community. Chichester: Ellis Horwood. Cruttenden, A. (1994) Gimson’s Pronunciation of English (5th edn.). London: Arnold. 275

Crystal, D. (1995) The Cambridge Encyclopedia of the English Language. Cambridge: Cambridge University Press. Department of the Environment, Transport and the Regions (1996) Urban Trends in England: Evidence from 1991 Census. Urban Research Summary No.

8. Docherty, G.J. & P. Foulkes (1999) Derby and Newcastle: instrumental phonetics and variationist studies. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp47- 71. Docherty, G.J., P. Foulkes, J. Milroy, L. Milroy & D. Walshaw (1997) Descriptive adequacy in phonology: a variationist perspective. Journal o f Linguistics 33: 275-31. Docherty, G.J. & P. Foulkes (in press) Variability in (r) production - instrumental perspectives. In H. Van de Velde & R. van Hout (eds.) r- atics: Sociolinguistic, Phonetic and Phonological Characteristics of /r/. Brussels: ILVP. Dorian, N.C. (1982) Defining the speech community to include its working margins. In S. Romaine (ed.) Sociolinguistic Variation in Speech Communities. London: Arnold, pp25-34. Douglas-Cowie, E. (1978) Linguistic code-switching in a Northern Irish village: social interaction and social ambition. In P. Trudgill (ed.) Sociolinguistic Patterns in British English. London: Edward Arnold, pp37-51. Downes, W. (1984) Language and Society. London: Fontana. Dyer, J.A. (2000) Language and identity in a Scottish-English community: a phonological and discoursal analysis. Unpublished PhD. Dissertation, University of Michigan. Dyer, J.A. (in press) ‘We all speak the same round here.’ Dialect levelling in a Scottish English community. To appear in Journal of Sociolinguistics

6(2). Eckert, P. (1989) The Whole Woman: Sex and Gender Differences in Variation. Language Variation and Change 1: 245-67. Eckert, P. (1997) The Whole Woman: Sex and Gender Differences in Variation. In N. Coupland & A. Jaworski (eds.) Sociolinguistics: A Reader and Coursebook. London: Macmillan Press Ltd., pp212-228. 276

Eckert, P. (1998) Age as a Sociolinguistic Variable. In F. Coulmas (ed.) The Handbook of Sociolinguistics. Oxford: Blackwell, ppl 51-167. Eckert, P. (2000) Linguistic Variation as Social Practice: The Linguistic Construction o f Identity in Belten High. Oxford: Blackwell. Edwards, V.K., P. Trudgill & B. Weltens (1984) The grammar of English dialect: a survey of research: a report to the ESRC Education and Human Development Committee. London: Economic and Social Research Council. Fabricius, A.H. (2000) T-glottalling between stigma and prestige: a sociolinguistic study of modem RP. Unpublished PhD. Dissertation, Copenhagen Business School. Foulkes, P. (1997) English [r]-sandhi: a sociolinguistic perspective. Histoire Epistemologie Langage 19(1): 73-96. Foulkes, P. & G. J. Docherty (1999a) Urban Voices - Overview. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, p p l-24. Foulkes, P. & G.J. Docherty (1999b) Urban Voices: Accent Studies in the British Isles. London: Arnold. Foulkes, P. & G.J. Docherty (2000) Another chapter in the story of /r/: ‘Labiodental’ variants in British English. Journal of Sociolinguistics 4(1): 30-59. Foulkes, P. & G.J. Docherty (in press). Variation and change in British English (r). In H. Van de Velde & R. van Hout (eds.) r-atics: Sociolinguistic, Phonetic and Phonological Characteristics of/r/ Brussels: ILVP. Foulkes, P., G.J. Docherty & D. Watt (1999) Tracking the Emergence of Structured Variation - Realisations of (t) by Newcastle Children. Leeds Working Papers in Linguistics and Phonetics vol. 7. Department of Linguistics and Phonetics, University of Leeds, ppl-25. Freed, A. (1996) Language and gender in an experimental setting. In V. Bergvall, J. Bing & A. Freed (eds.) Rethinking Language and Gender Research: Theory and Practice. New York: Longman, pp77-97. Fromkin, V. (1985) Phonetic Linguistics. Orlando: Academic Press. Gairns, R. & S. Redman (1986) Working with words: A guide to teaching and learning vocabulary. Cambridge: Cambridge University Press. 277

Gerritsen, M. (1987) Sociolinguistic developments as a diffusion process. In U. Ammon, N. Dittmar & K. Mattheier (eds.) Sociolinguistics: an international handbook of the science of language and society. Berlin: De Gruyter, ppl574-1591. Giddens, A. (1989) Sociology. Cambridge: Polity. Giegerich, H.J. (1992) English Phonology: An introduction. Cambridge: Cambridge University Press. Giles, H. & P.F. Powseland (1975) Speech Style and Social Evaluation. London: Academic Press. Giles, H. & P. Smith (1979) Accommodation theory: optimal levels of convergence. In H. Giles & R. St Clair (eds.) Language and Social Psychology. Oxford: Blackwell, pp45-65. Gimson, A.C. (1980) An Introduction to the Pronunciation of English (3 rd edn.). London: Arnold. Gimson, A.C. (1989) An Introduction to the Pronunciation o f English (4th edn.). London: Arnold. Gove, W.R. (1980) The Labelling of Deviance: evaluating a perspective. California: Sage. Grant, W. (1913) The pronunciation of English in Scotland. Cambridge: Cambridge University Press. Gumperz, J.J. (1968) The speech community. In International Encyclopedia of the Social Sciences. London: Macmillan, pp381-386. Gumperz, J.J. (1982) Language and Social Identity. Cambridge: Cambridge University Press. Guy, G. (1980) Variation in the group and the individual: the case of final stop deletion. In W. Labov (ed.) Locating language in time and space. New York: Academic Press, ppl-36. Hernandez Campoy, J. (2000) Requisitos teorico-metodologicos para el estudio geolinguistico del dialecto Murciano. In J.M. Jimenez Cano (ed.) Estudios sociolinguisticos del dialecto Murciano. Murcia: Universidad de Murcia. Hirst, T.O. (1914) The popular English of Lancaster and district. Le maitre phonetique 26-29. 278

Holmes, J. (1992) An Introduction to Sociolinguistics. Harlow: Addison Wesley Longman Limited. Holmes, J. (1995) Women, Men and Politeness. NewYork: Longman. Houk, C.L. (1968) Methodology of an urban speech survey. Leeds Studies in English, University of Leeds, ppl 15-128. Hudson, R.A. (1980) Sociolinguistics. Cambridge: Cambridge University Press. Hughes, A. & P. Trudgill (1996) English Accents and Dialects. (3rd edn.). London: Arnold. Irvine, J.T. & S. Gal (2000) Language ideology and linguistic differentiation. In P. Kroskrity (ed.) Regimes of Language. Santa Fe: School of American Research Press, pp35-83. Jerspersen, O. (1922) Language: Its Nature, Development and Origin. London: George Allen & Unwin Ltd. Johnston, P.A. (1985a) Linguistic atlases and sociolinguistics. In J.M. Kirk, S. Sanderson & J.D.A. Widdowson (eds.) Studies in Linguistic Geography. Beckenham: Croon Helm, pp81-93. Johnston, P.A. (1985b) The rise and fall of the Morningside/Kelvinside accent. In M. Gorlach (ed.) Focus on Scotland. Amsterdam: John Benjamins, pp37-56. Johnstone, B. (1999) Place, Globalization, and Linguistic Variation. Paper presented at Methods in Sociolinguistics, Pitzer College, Claremont, November 1999. Johnstone, B. & M. Bean (1997) Self-Expression and linguistic variation. Language in Society 26: 221-246. Jones, D. (1909) The pronunciation of English. Cambridge: Cambridge University Press. Kellett, A. & I. Dewhirst (1997) A Century of Yorkshire Dialect Selections from the Transactions o f the Yorkshire Dialect Society. Otley: Smith Settle. Kerswill, P. (1987) Levels of linguistic variation in Durham. Journal of Linguistics 23: 25-49. Kerswill, P. (1996) Children, adolescents and language change. Language Variation and Change 8: 177-202. Kerswill, P. (2000) Salience. Paper presented at Department of Linguistics, Queen Mary & Westfield College, 8 February 2000. 279

Kerswill, P. & A. Williams (1994) A new dialect in a new city: children’s and adult’s speech in Milton Keynes. Final report presented to the Economic and Social Research Council, July 1994. Kerswill, P. & A. Williams (1997) Creating a new town koine: children and language change in Milton Keynes. Reading Working Papers in Linguistics 3. Department of Linguistic Science, University of Reading, pp205-257. Kerswill, P. & A. Williams (2000) Dialect recognition and speech community focusing in new and old towns in England: the effects of dialect levelling, demography and social networks. In D. Long & D. Preston (eds.) A Handbook of Perceptual Dialectology Vol. 2. Amsterdam: Benjamins. Kiely, R., D. McCrone, F. Bechhofer & S. Stewart (2000) Debatable Land: national and local identity in a border town. Sociological Research Online 5(2). URL: Kretzschmar, Jr., W.A., V.G. McDavid, T.K. Lerud & E. Johnson (1994) Handbook of the Linguistic Atlas of the Middle and South Atlantic States. Chicago: University of Chicago Press. Labov, W. (1963) The social motivation of a sound change. Word 19: 273-309. Labov, W. (1966) The Social Stratification of English in New York City. Washington D.C.: Center for Applied Linguistics. Labov, W. (1972) Sociolinguistic Patterns. Philadelphia: University of Philadelphia Press. Labov, W. (1979) Locating the frontier between social and psychological factors in linguistic variation. In C.J. Fillmore, D. Kempler & W.S.-Y. Wang (eds.) Individual differences in language ability and language behavior. New York: Academic Press, pp327-339. Labov, W. (1990) The intersection of sex and social class in the course of linguistic change. Language Variation and Change 2: 205-54. Labov, W. (1994) Principles o f Linguistic Change. Volume 1: Internal Factors. Oxford: Blackwell. Labov, W. (2001) Principles o f Linguistic Change. Volume 2: Social Factors. Oxford: Blackwell. Labov, W. (forthcoming) Principles o f Linguistic Change. Volume 3: Cognitive Factors. 280

Lakoff, R. (1975) Language and Woman’s Place. New York: Harper & Row. Ladefoged, P. & I. Maddieson (1996) The Sounds o f the World’s Languages. Oxford: Blackwell. Lehrer, A. (1974) Semantic Fields and Lexical Structure. Amsterdam: North- Holland Linguistic Series. Le Page, R.B. & A. Tabouret-Keller (1985) Acts of Identity: creole-based approaches to language and ethnicity. Cambridge: Cambridge University Press. Lewis, R. (1996) The Evolution of a Political Culture: Middlesbrough 1850- 1950. In A.J. Pollard (ed.) Middlesbrough: Town and Community 1830- 1950. Stroud: Sutton Publishing Ltd., ppl03-126. Lewis, R. & D. Ward (1995) Culture, Politics and Assimilation: The Welsh on Teesside, c. 1850-1940. The Welsh History Review 17(4): 550-570.

Lindau, M. (1985) The story of It/. In V. Fromkin (ed.) Phonetic Linguistics. Orlando: Academic Press, ppl57-168. Lippi-Green, R. (1997) English with an Accent: Language, ideology, and discrimination in the United States. London: Routledge. Llamas, C. (1998) Language variation and innovation in Middlesbrough: a pilot study. Leeds Working Papers in Linguistics vol. 6. Department of Linguistics and Phonetics, University of Leeds, pp97-l 14. Local, J.K., J. Kelly & W.H.G. Wells (1986) Towards a Phonology of Conversation: Turn-taking in Tyneside English. Journal of Linguistics 22(2): 411-437. Lodge, K.R. (1984) Studies in the Phonology of Colloquial English. Croom Helm: Beckenham. Lyons, J. (1995) Linguistic Semantics: an Introduction. Cambridge: Cambridge University Press. Macafee, C. (1983) Glasgow. Amsterdam: Benjamins. Macafee, C. (1994) Traditional Dialect in the Modern World: A Glasgow Case Study. Frankfurt: Peter Lang. Macafee, C. (1997) Ongoing change in modem Scots: the social dimension. In C. Jones (ed.) The Edinburgh History of the Scots Language. Edinburgh: Edinburgh University Press, pp514-548. 281

Macaulay, R.K.S. (1978) Variation and consistency in Glaswegian English. In P. Trudgill (ed.) Sociolinguistic Patterns of British English. London: Edward Arnold, pp 132-143. Machyn, H. (1848) The Diary o f Henry Machyn, Citizen and Merchant-Taylor o f London: from A.D. 1550 to A.D. 1563. J G. Nichols (ed.). London: The Camden Society. Maddieson, I. (1984) Patterns o f sounds. Cambridge: Cambridge University Press. Maltz, D. & R. Borker (1982) A cultural approach to male-female miscommunication. In J. Gumperz (ed.) Language and Social Identity. Cambridge: Cambridge University Press, ppl96-216. Marshall, J. (2001) /t/ glottaling in the north of Scotland. Paper presented at The Third UK Language Variation and Change Conference, University of York, July 2001. Mather, J.Y., H.H. Speitel & G.W. Leslie (1975) The Linguistic Atlas of Scotland. 3 vols. London: Croom Helm. Mathisen, A.G. (1999) Sandwell, West Midlands: ambiguous perspectives on gender patterns and models of change. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, ppl07-123. Matthews, W. (1938) Cockney past and present. London: Kegan Paul. McCafferty, K. (1999) (London)Derry: between Ulster and local speech - class, ethnicity and language change. In P. Foulkes & G.J Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp246-264. McCawley, J.D. (1971) Tense and time reference in English. In C.J. Filmore & D.T. Langedoen (eds.) Studies in linguistic semantics. New York: Holt, Rinehart and Winston, pp96-103. McMahon, A. (1994) Understanding language change. Cambridge: Cambridge University Press. Mees, I. (1987) Glottal stop as a prestigious feature in Cardiff English. English World Wide 8(1): 25-39. 282

Mees, I.M. & B. Collins (1999) Cardiff: a real-time study of glottalization. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp85-202. Miller, G.A. & P.E. Nicely (1955) Analysis of perceptual confusions among some English consonants. Journal of the Acoustical Society of America 27: 338-53. Milroy, J. (1992) Linguistic Variation and Change: On the Historical Sociolinguistics of English. Oxford: Blackwell. Milroy, J. (1996) A current change in British English: variation in (th) in Derby. Newcastle and Durham Working Papers in Linguistics 4: 213-222. Milroy, L. (1987a) Language and Social Networks (2nd edn.). Oxford: Blackwell. Milroy, L. (1987b) Observing and Analysing Natural Language. A Critical Account of Sociolinguistic Method. Oxford: Blackwell. Milroy, L. (1997) Gender, social class and supra-local norms: dialect levelling as language change. Paper presented at SALSA, 13 April 1997. Milroy, L. (2000) Two nations divided by the same language (and different language ideologies). Journal of Linguistic Anthropology 9(1). Milroy, L. (2001) Internally and externally motivated language change: Assessing the effects of “nature” and “nurture”. Keynote address delivered at the Third UK Language Variation and Change Conference, University of York, July 2001. Milroy, L., J. Milroy & G.J. Docherty (1997) Phonological Variation and Change in Contemporary Spoken British English. Final Report to the UK Economic and Social Research Council. Milroy, J., L. Milroy & S. Hartley (1994) Local and supra-local change in British English: The case of glottalisation. English World Wide 15(1): 1-33. Moorsom, N. (1993) A Journey through the History of Middlesbrough. Middlesbrough: Gordian Print. Moorsom, N. (1995) Highlights in the History o f Cleveland. Middlesbrough: Cleveland County Council. Moorsom, N. (1996) Middlesbrough Re-Born: The Evolution of a Local Authority. Middlesbrough: Moorsom. Mullany, L. (2000) The application of current language and gender theory to managerial meeting discourse. Nottingham Linguistic Circular 15:1-13. 283

Murray, A.V. (forthcoming) Labialisation of non-postvocalic /r/ in varieties of south-eastern English. Nicholson, T. (1996) ‘Jacky’ and the Jubilee: Middlesbrough’s Creation Myth. In A.J. Pollard (ed.) Middlesbrough: Town and Community 1830-1950. Stroud: Sutton Publishing Ltd., pp32-52. Newbrook, M. (1986) Sociolinguistic Reflexes of Dialect Interference in West Wirral. Bern & Frankfurt am Main: Peter Lang Verlag. Newbrook, M. (1999) West Wirral: norms, self-reports and usage. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp90-106. Nolan, F.J. & T. Oh (1996) Identical twins, different voices. Journal of Forensic Linguistics 3: 39-49. O’Connor, J.D. (1973) Phonetics. London: Penguin. Ofitedal, M. (1975) The Phonemic Analysis of Scottish Gaelic. Phonetica 32(2): 130-140. Ohala, J.J. (1993) The phonetics of sound change. In C. Jones (ed.) Historical " Linguistics: Problems and Perspectives. London: Longman, pp237-278. One North East: The Development Agency for the North East of England (1999) Regional Economic Strategy for the North East: Unlocking Our Potential. Orton, H. & E. Dieth (1962-1971) Survey of English Dialects. Introduction and 4 vols. in 3 parts. Leeds: E. J. Arnold. Orton, H., S. Sanderson & J.D.A. Widdowson (1978) The Linguistic Atlas of England. London: Croom Helm. Parry, D. (1977-1979) The Survey of Anglo-Welsh Dialects. 2 vols. Swansea: University College of Swansea. Payne, A.C. (1980) Factors controlling the acquisition of the Philadelphia dialect by out-of-state children. In W. Labov (ed.) Locating Language in Time and Space. New York: Academic Press, ppl43-178. Pollard, A.J. (1996) Middlesbrough: Town and Community 1830-1950. Stroud: Sutton Publishing Ltd. Pratt, T.K. (1983) A case for direct questioning in traditional fieldwork. American Speech 58: 150-55. Preston, D.R. (1989) Perceptual Dialectology. Dordrecht: Foris. 284

Reid, E. (1978) Social and stylistic variation in the speech of children: some evidence from Edinburgh. In P. Trudgill (ed.) Sociolinguistic Patterns in British English. London: Edward Arnold, pp 158-173. Richardson, J. & G. Maxwell (1998) An Audit of Crime and Disorder in Middlesbrough: A Consultation Report. Middlesbrough Council. Roberts, J. & W. Labov (1995) Learning to talk Philadelphian. Language Variation and Change 9: 249-266. Romaine, S. (1975) Linguistics variability in the speech of some Edinburgh schoolchildren. MLitt thesis, University of Edinburgh. Romaine, S. (1980) A critical overview of the methodology of urban British sociolinguistics. English World Wide 1: 163-198. Romaine, S. (1982) What is a speech community? In S. Romaine (ed.) Sociolinguistic Variation in Speech Communities. London: Arnold, ppl3- 24. Romaine, S. (1984) The Language of Children and Adolescents: The Acquisition of Communicative Competence. Oxford: Blackwell. Rosewame, D. (1984) Estuary English. The Times Educational Supplement. 19th October. Rosewame, D. (1994) Estuary English: tomorrow’s RP? English Today 10: 3-8. Sacks, H., E. Schegloff & G. Jefferson (1974) A simplest systematics for the organisation of turn-taking in conversation. Language 50: 696-735. Sankoff, G. (1980) The Social Life o f Language. Philadelphia: University of Pennsylvania Press. Schegloff, E. (2000) Overlapping talk and the organization of turn-taking in Conversation. Language in Society 29(1): 1-63. Shuken, C. (1984) Highland and Island English. In P. Trudgill (ed.) Language in the British Isles. Cambridge: Cambridge University Press, ppl52-177. Siegel, J. (1985) Koines and koineization. Language in Society 14: 357-378. Silverstein, M. (1979) Language structure and linguistic ideology. In R. Clyne, W. Hanks & C. Hofbauer (eds.) The Elements: A parasession on linguistic units and levels. Chicago: Chicago Linguistic Society, ppl93- 247. Silverstein, M. (1992) The uses and utility of ideology: some reflections. Pragmatics 2(3): 311-323. 285

Silverstein, M. (1995) Indexical order and the dialectics of social life. Keynote address delivered at SALSA III, University of Austin, Texas. Spender, D. (1980) Man Made Language. London: Routledge. Stoddart, J., C. Upton & J.D.A. Widdowson (1999) Sheffield dialect in the 1990s: revisiting the concept of NORMs. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp72-89. Stuart-Smith, J. (1999) Glasgow: accent and voice quality. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp203-222. Stuart-Smith, J. (2001) Should variationists worry about the influence of TV on speech? Keynote address delivered at the Third UK Language Variation and Change Conference, University of York, July 2001. Stuart-Smith, J. & C. Tweedie (2000) Accent change in Glasgow: A sociophonetic investigation. Final report to the Leverhulme Trust. Sweet, H. (1874) History of English Sounds. London: Triibner. Sweet, H. (1908) The sounds o f English. Oxford: Oxford University Press. Taylor, D. (1996) The Infant Hercules and the Augean Stables: a Century of Economic and Social Development in Middlesbrough, c. 1840-1939. In A.J. Pollard (ed.) Middlesbrough: Town and Community 1830-1950. Stroud: Sutton Publishing Ltd., pp53-80. Teesside Training and Enterprise Council (TEC) (1997) People and Jobs in Teesside: A Guide to the Local Labour Market, Middlesbrough. Tees Valley Training and Enterprise Council (TEC) (2001) The Tees Valley Labour Market: A Summary of Key Information, Middlesbrough. The Observer (December 3, 2000) Tweenies: what little girls are made of. The Sunday Times (August 9, 1998) Wild, wild women. Todd, G. (1987) Geordie words and phrases: an aid to communication on Tyneside and thereabouts. Northumberland: Butler. Thrainsson, H. (1978) On the Phonology of Icelandic Preaspiration. Nordic Journal of Linguistics 1(1): 3-54. Tollfree, L. (1999) South East London English: discrete versus continuous modelling of consonantal reduction. In P. Foulkes. & G.J. Docherty (eds.) 286

Urban Voices: Accent Studies in the British Isles. London: Arnold, p p l63-184. Torgersen, E.N. (1997) Some phonological innovations in southeastern British English. Cand. Philo 1. Dissertation, University of Bergen. Torgersen, E.N. & P. Kerswill (2001) Acoustic measurements of some vowels in south-eastern British English - further evidence for chain shift and accent levelling. Paper presented at The Third UK Language Variation and Change Conference, University of York, July 2001. Trousdale, G.M. (2000) Variation and (Socio)linguistic Theory: a case study o f Tyneside English. Unpublished PhD. Dissertation, University of Edinburgh. Trudgill, P (1974) The Social Differentiation of English in Norwich. Cambridge: Cambridge University Press. Trudgill, P. (1983a) On Dialect: Social and Geographical Perspectives. Oxford: Blackwell. Trudgill, P. (1983b) Acts of conflicting identity: The sociolinguistics of British pop-song pronunciation. In P. Trudgill On Dialect: Social and Geographical Perspectives. Oxford: Blackwell. Trudgill, P. (1986) Dialects in Contact. Oxford: Blackwell. Trudgill, P (1988) Norwich revisited. English World Wide 9: 33-49. Trudgill, P. (1990) The Dialects of England. Oxford: Blackwell. Trudgill, P. (1997) The Social Differentiation of English in Norwich. In N. Coupland & A. Jaworski (eds.) Sociolinguistics: A Reader and Coursebook. London: Macmillan Press Ltd., ppl79-184. Trudgill, P. (1999b) A window on the past: “colonial lag” and New Zealand evidence for the phonology of 19th century English. Paper presented at 10th International Conference on Methods in Dialectology, Newfoundland, August 1999. Trudgill, P. (1999a) Norwich: endogenous and exogenous linguistic change. In P. Foulkes, & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, ppl24-140. Trudgill, P. (2000) Sociolinguistics: and Introduction to language and society (4th edn.). London: Penguin. 287

Turner, J. (1996) The Frontier Revisited: Thrift and Fellowship in the New Industrial Town, c. 1830-1914. In A.J. Pollard (ed.) Middlesbrough: Town and Community 1830-1950. Stroud: Sutton Publishing Ltd., pp81-

102. Ullman, S. (1957) The Principles of Semantics (2nd edn.) Oxford: Blackwell. Underwood, G.N. (1988) Accent and Identity. In A.R. Thomas (ed.) Methods in Dialectology. Clevedon: Multilingual Matters, pp406-427. Upton, C. & J.D.A. Widdowson (1996) An Atlas of English Dialects. Oxford: Oxford University Press. Ur, P. (1988) Grammar Practice Activities: A Practical Guide for Teachers. Cambridge: Cambridge University Press. Wakelin, M.F. (1972) Patterns in the Folk Speech o f the British Isles. London: Athlone Press. Wardhaugh R. (1992) An Introduction to Sociolinguistics (2nd edn.). Oxford: Blackwell. Watt, D.J.L. (1998) Variation and Change in the Vowel System o f Tyneside English. Unpublished PhD. Dissertation, University of Newcastle. Watt, D.J.L. (2000a) Phonetic parallels between the close-mid vowels of Tyneside English: Are they internally or externally motivated? Language Variation and Change 12: 69-101. Watt, D.J.L. (2000b) An acoustic investigation of vowel fronting in urban Yorkshire English. Paper presented at The First International Conference on Language Variation in Europe, Barcelona, June 2000. Watt, D.J.L. (in press) ‘I don’t speak with a Geordie accent, I speak, like, the Northern accent’: contact-induced levelling in the Tyneside vowel system. To appear in Journal of Sociolinguistics 6(2). Watt, D.J.L. & C. Ingham (2000) Durational Evidence of the Scottish Vowel Length Rule in Berwick English. Leeds Working Papers in Linguistics and Phonetics 8, Department of Linguistics and Phonetics, University of Leeds, pp.205-226 Watt, D.J.L. & L. Milroy (1999) Patterns of variation and change in three Tyneside vowels: Is this dialect levelling? In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, pp25-46. 288

Watt, D.J.L. & J. Tillotson (1999) A spectrographic analysis of vowel fronting in Bradford English. Leeds Working Papers in Linguistics and Phonetics 7, Department of Linguistics and Phonetics, University of Leeds, pp206- 234. Weinreich, U., W. Labov & M.I. Herzog (1968) A Theory of Language Change. In W.P. Lehmann & Y. Malkiel (eds.) Directions for Historical Linguistics: A Symposium. Austin: University of Texas Press, pp95-189. Wells, J.C. (1982) Accents of English (3 vols.). Cambridge: Cambridge University Press. Wells, J.C. (1990) A phonetic update on RP. Moderna Sprak. LXXXII(l): 3.9. Wells, J. C. (1994) The Cockneyfication of RP? In G. Melchers & N.-L. Johannesson (eds.) Nonstandard varieties of language. Stockholm: Almqvist & Wiksell, ppl98-205. Wells, J.C. (1997) What’s happening to Received Pronunciation? English Phonetics (English Phonetic Society of Japan). 1:13-23. Williams, A. & P. Kerswill (1999) Dialect levelling: continuity vs. change in Milton Keynes, Reading and Hull. In P. Foulkes & G.J. Docherty (eds.) Urban Voices: Accent Studies in the British Isles. London: Arnold, ppl41-162. Wolfson, N. (1982) CHP: the conversational historic present in American English narrative. Dordrecht: Foris. Woolard, K.A. (1992) Language ideology: Issues and approaches. Pragmatics 2(3): 235-249. Woolard, K.A. & B. Schieffelin (1994) Language Ideology. Annual Review of Anthropology 23: 55-82. Wright, J. (1892) A Grammar o f Windhill in the West Riding o f Yorkshire. London: Kegan Paul. Wright, P. (1981) Cockney Dialect and Slang. London: B. T. Batsford. Zimmerman, D. & C. West (1975) Sex-roles, interruptions and silences in conversation. In B. Thome, C. Kramarae & N. Henley (eds.) Language, Gender and Society. Rowley: Newbury House, pp89-101.